CASES
Each entry is one engagement — what it was, and the part AEON played in it. Most entries close with a note on where that field is heading. The deliverables themselves are not published.
Not everything AEON has done can sit on one page, so this is a selection, weighted towards the last ten years or so.
Sea-Sword — Promotional Film
Promotional Film
Sea-Sword — Promotional Film
Sea-Sword is LIG D&A's family of maritime unmanned systems, covering surface, aerial and underwater vehicles.
Sea-Sword is a family of maritime unmanned systems developed by LIG D&A, the South Korean defence company formerly known as LIG Nex1. The family brings together unmanned surface, aerial and underwater vehicles, among them mine countermeasure craft and one-way attack unmanned surface vessels, built on modular lines and intended to be operated in numbers.
AEON planned and produced the promotional film for the Sea-Sword family.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Few areas of military technology have moved as quickly since 2022 as uncrewed maritime systems. The Black Sea is the reason. Ukraine's Sea Baby and Magura boats began as remotely piloted explosive craft, and they now act as carriers in their own right: Sea Baby can launch small FPV drones from side compartments, and it can continue autonomously when its link is jammed. The technique is also being exported. In June 2026 US special forces used a Ukrainian Magura to sink a target ship during Balikatan 2026 off the Philippines, the first time the boat had been used in the Indo-Pacific.
Large navies are now trying to turn that improvisation into programmes. The US Navy has changed its approach more than once. In March 2026 it replaced the Modular Attack Surface Craft effort with a Medium Unmanned Surface Vessel family of systems, run as an open marketplace for production-ready designs. Each vessel is to carry at least two 40-foot container-sized payloads that can be swapped between missions, and the first production boats are due in fiscal 2027. The shift came after the Pentagon's Replicator initiative, launched in August 2023, set out to field thousands of attritable systems and selected uncrewed surface vehicles of several sizes alongside undersea craft, though by the August 2025 deadline it had fielded hundreds rather than thousands.
Below the surface, the trend is the same. Australia moved straight from a co-development programme to a five-year production contract for Anduril's Ghost Shark extra-large autonomous undersea vehicle in September 2025. The vehicle is built for long-range surveillance and strike, and from the start it was designed to carry interchangeable mission payloads and to be built in quantity.
Three patterns run through these programmes. First, navies are buying families rather than single hulls: surface, aerial and undersea vehicles that share control software and payload interfaces. Second, the payload has become the unit of capability, and the hull is increasingly a carrier for containerised sensors, effectors or decoys. Third, the systems are planned for numbers, which puts weight on the shore side: launch and recovery, charging, maintenance and data handling all become part of the weapon system, not just support for it.
For operators, the change is from steering a boat to supervising a group of them. One crew may be responsible for many vessels spread across a wide area, each with its own payload, link quality and level of autonomy. The questions that matter are how a supervisor sees which vehicles need attention, how authority passes between human and machine when a link drops, and how a harbour team and a mission team share one picture of the fleet. Interfaces built around exceptions, clear indications of autonomy state and handover will matter as much as the vessels themselves.
MR-based Swarm Drone Operation & Control
MR UX Design
MR-based Swarm Drone Operation & Control

MR-based Swarm Drone Operation & Control is a mixed-reality interface for directing a swarm of drones as a coordinated group.
A drone swarm is a group of unmanned aircraft that coordinate with one another, so that they can be tasked as a single formation rather than flown one by one. Mixed reality (MR) combines a view of the real surroundings with computer-generated information, usually through a head-mounted display, and is one of several approaches being explored to let a small number of people supervise many vehicles.
AEON carried out UX design for LIG D&A's mixed-reality interface for operating and controlling a drone swarm.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The central problem of swarm warfare was stated clearly by DARPA's OFFSET programme. Its aim was to let a small infantry unit use swarms of up to 250 air and ground robots in a city, and its premise was that nobody wants a platoon of soldiers each holding a joystick. At the final OFFSET field experiment at Fort Campbell in November 2021, more than a hundred robots mapped an urban site and located a target, directed by a single operator working with a swarm commander through tablet and virtual-reality interfaces.
What was an experiment in 2021 is now being tested with live weapons. In January 2026, at a range event at Camp Blanding in Florida, a single operator using Auterion's Nemyx swarm software assigned three strike drones to three separate targets. The human set the mission parameters and authorised the strike, and the software handled navigation, formation, deconfliction and terminal guidance. The test was an early event in the Pentagon's Swarm Forge effort, which pairs military units with commercial developers to iterate on autonomous tactics.
The display side has proved harder than the autonomy. The US Army's Integrated Visual Augmentation System, a HoloLens-derived mixed-reality headset, was meant to give every soldier a shared digital overlay. In July 2026 the Government Accountability Office reported that none of the roughly 10,000 IVAS units built would be fielded, after users reported headaches, eyestrain, nausea, poor low-light performance and added helmet weight. The effort has become Soldier Borne Mission Command, a rapid prototyping programme led by Anduril. For Project Convergence in summer 2026, L3Harris and Anduril planned to pair the ENVG-B night-vision goggle with Anduril's Lattice networking software, an approach that starts from a display soldiers already wear rather than a new headset.
Taken together, these programmes point the same way. Control is moving from piloting individual vehicles to stating intent for a group: an area to search, a target set, constraints and a time window, with the software working out who does what. Mixed and augmented reality appeal because a swarm is a spatial object, and placing its tasks and tracks on the operator's own view of the terrain avoids constant translation between a map and the world outside. But the IVAS experience shows that the wearable has to survive the field before any interface on it matters.
For the people involved, the main questions are about attention and authority. An operator supervising dozens of vehicles cannot watch each one, so the interface has to bring forward the few events that need a decision and keep the rest in the background. Weapons release must stay visibly under human control, and the operator needs to see at a glance what the swarm is doing on its own. Comfort, low-light legibility and weight are not secondary concerns; they decide whether the system gets worn at all.

SoS Presentation Materials

A System of Systems (SoS) is an approach in which independent platforms, sensors and weapons are networked so that they operate as one larger capability; in naval warfare it underpins concepts for distributed and manned-unmanned operations.
In naval operations, a System of Systems links ships, submarines, aircraft, unmanned vehicles and shore-based nodes through common networks and shared data, so that a sensor on one platform can cue a weapon or another sensor elsewhere. The approach is particularly relevant to anti-submarine warfare, where detection depends on combining sonars, sonobuoys, maritime patrol aircraft and helicopters over wide sea areas. The Republic of Korea Navy is pursuing it through its manned-unmanned combined combat concept, known as Navy Sea GHOST.
AEON planned and produced, as video and multimedia, presentation material for a Republic of Korea Navy academic conference proposing a System of Systems concept for naval operations, including anti-submarine warfare.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The idea that a navy fights as a network rather than as a collection of ships is not new, but over the past decade it has hardened into doctrine. The US Navy's Distributed Maritime Operations concept spreads weapons and sensors across many more platforms so that the loss of any one does not break the whole. A Congressional Research Service report updated in January 2026 describes planned large unmanned surface vessels as adjunct weapon magazines and sensor platforms intended to help implement that concept, a reminder that the system of systems is now being built partly from hulls with no crew aboard.
Distribution only works if the pieces can talk to each other. Project Overmatch is the US Navy's effort to provide the networks, data architecture and tools that tie ships, aircraft and weapons together and connect them to the wider joint force. In 2024 the commander of US Third Fleet said Overmatch capabilities were already fielded on three carrier strike groups. The hard problem is less the radio link than the data: shared tracks, common formats, and software that can be updated at the pace of the threat.
South Korea is following a similar path at its own scale. The Republic of Korea Navy's manned-unmanned concept, Navy Sea GHOST, is moving from concept art to hardware. In late 2025 a consortium of HD Hyundai Heavy Industries, HJ Heavy Industries and LIG Nex1 agreed with ADD to build a verification platform for combat USV control and autonomy technologies. Navy plans discussed in 2025 envisage adapting the Dokdo-class amphibious ships for drone operations and building a dedicated drone carrier for service in the late 2030s.
The most telling recent demonstration came in August 2026, when Hanwha Systems and the navy ran a manned-unmanned mine countermeasures exercise at Busan. An AI-based command system turned natural-language instructions into a search plan and tasked unmanned surface and underwater vehicles, a crewed minesweeper and a UAV, with satellite links sharing the picture. Mine warfare and anti-submarine warfare share the same character: wide areas, faint signals and a premium on persistence, which is exactly where many distributed sensors can outperform a few exquisite platforms.
For operators, this shift changes the nature of command. A watch officer who once managed a single ship's sensors now supervises a web of crewed and uncrewed nodes, some far over the horizon and some acting with partial autonomy. The interface has to show not only where objects of interest are, but how confident the network is about them, which node supplied each piece of evidence, and what each unmanned asset is currently permitted to do. As AI tools begin to propose plans, the human task becomes judging and adjusting those plans quickly, and that depends on designs that make machine reasoning legible without burying the operator in detail.
GR-LIG Multi-Robot Control
Service Design & UX
GR-LIG Multi-Robot Control

GR-LIG Multi-Robot Control is a control package for operating several unmanned ground robots, including Ghost Robotics quadrupeds, as one team.
Ghost Robotics is a Philadelphia-based maker of legged robots, best known for the Vision 60 quadrupedal unmanned ground vehicle. LIG Nex1, now LIG D&A, completed the acquisition of a controlling 60 per cent stake in the company in July 2024. Multi-robot control refers to the software and procedures that let operators task and monitor several robots together rather than one at a time.
AEON provided service design and UX for a multi-robot control package spanning the control room, field operators and payload suppliers.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Legged robots entered military service through the least dramatic door: base security. In March 2021 Tyndall Air Force Base in Florida received quadrupedal unmanned ground vehicles from Ghost Robotics and Immersive Wisdom, described as the first of their kind integrated onto a military installation. They were used as mobile sensors that could patrol remote parts of the base, so that human defenders could concentrate elsewhere.
Since then, experiments have moved closer to combat. US Marine Corps special operators have tested Vision 60 robots fitted with Onyx's SENTRY remote weapon system, which detects targets and alerts a human who decides whether to fire. The command stressed that it was not fielding the capability and that weapons are only one possible payload alongside sensors and electronic warfare; the robots were also reported to be doing tunnel work and perimeter security. In August 2024 the US Army's Operation Hard Kill at Fort Drum showed a Vision 60 carrying a carbine on a small turret as part of a counter-drone layer, operated from tablet-like controllers.
Other armies are on a similar path. South Korea's procurement agency announced in August 2024 that a four-legged counterterrorism robot developed with Hyundai Rotem and Rainbow Robotics would be trialled with Army special forces and a forward division, for search, reconnaissance and sentry duty. China drew attention the same year when its forces showed a rifle-carrying robot dog under remote control in the Golden Dragon exercise with Cambodia, presented as a member of urban combat teams.
The more important shift is from single robots to teams. The clearest public demonstration remains the DARPA Subterranean Challenge, which team CERBERUS won in September 2021 with four ANYmal C legged robots working alongside other platforms to explore underground courses and report what they found. Military practice is heading the same way: legged robots for stairs, rubble and tunnels, drones for overwatch, and wheeled or tracked vehicles to carry them forward. The robots are increasingly treated as payload carriers, with sensors, manipulators or effectors from different suppliers attached through common interfaces. That makes payload integration, software updates and shared autonomy as important as the robot's legs.
For operators, a robot team changes the job. Instead of driving one machine, a person allocates tasks across several and watches for the moments when one needs help. Responsibility is split between a control room with the wider picture and soldiers in the field who see what the robots see up close, and each needs to know what the other has authorised. Clear tasking by area or objective, readable status for every robot and smooth handover of control between locations decide whether a team of robots reduces workload or adds to it.
Global Multi-Layered Air Defence System — Promotional Film
Promotional Film
Global Multi-Layered Air Defence System — Promotional Film

LIG D&A's multi-layered air defence concept combines surface-to-air systems covering different ranges and altitudes, linked by a common command and control network.
Multi-layered air defence arranges several interceptor systems so that each covers a different band of range and altitude, from short-range defence up to upper-tier ballistic missile defence, with command and control linking sensors and shooters across the layers. LIG D&A, formerly LIG Nex1, is a South Korean defence company whose air defence work spans several of these layers, including the Cheongung-II medium-range system and the L-SAM long-range system developed with the Agency for Defense Development.
AEON planned and produced a promotional film presenting the company's multi-layered air defence concept.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since 2022, layered air defence has moved from briefing slides to daily practice. Two theatres in particular, Ukraine and the Middle East, have shown how layers behave under sustained pressure and what they cost to keep running.
Ukraine's experience has been one of continuous adaptation. Soviet-era S-300 and Buk systems carried the first months, NASAMS became operational in November 2022, and IRIS-T, Gepard guns and then Patriot followed, with Patriot intercepting Kinzhal missiles over Kyiv in 2023. As the Kyiv Post's review of four years of ground-based air defence describes, Ukraine also fitted Western missiles to Soviet launchers, formed mobile fire groups with guns and heavy machine guns to hunt Shahed drones and conserve missiles, and wove local observers into Western radars and weapons to build its own integrated network. Interceptor drones are the latest addition.
In the Middle East, the June 2025 war between Israel and Iran tested the upper layers at scale. According to a JINSA-republished analysis from February 2026, Israel and the United States tried to stop 322 Iranian ballistic missiles and intercepted 273, while THAAD batteries may have fired between a fifth and a half of the Pentagon's inventory. The Congressional Research Service cites one study estimating that THAAD accounted for almost half the interceptors used against Iranian medium-range missiles and that replenishment could take several years; THAAD was drawn on again during the 2026 regional campaign.
The common lesson is that layering is as much about magazine depth and cost as about coverage. Each layer exists partly to spare the one above it: guns, lasers and drones take the cheap threats, medium-range systems handle cruise missiles and shorter-range ballistic missiles, and scarce upper-tier interceptors are kept for what nothing else can reach. That allocation happens in command and control, which is why the connective tissue between layers has become as strategically important as the interceptors themselves.
This also changes how such architectures are explained. The classic diagram stacks layers as neat horizontal bands of range and altitude, but real coverage overlaps, threats move between bands, and the value of the architecture lies in coordination rather than in any single tier. Governments and manufacturers increasingly present air defence as an integrated family of sensors, shooters and C2 rather than as standalone products, and they rely on animation and film to show engagements unfolding across layers over time, something a static chart cannot convey.
For operators, layered defence means working inside a system larger than any one console. Crews need to know what the layers above and below them are doing, when a threat has been handed over and how many interceptors remain. Interfaces that make the whole architecture visible, and that make resource limits as plain as threat tracks, help people make sustainable decisions across long campaigns rather than only in single engagements.
M-SAM Block-III — Operations & Engagement Control Centre
UX Design & Prototype
M-SAM Block-III — Operations & Engagement Control Centre

M-SAM Block-III, also known as Cheongung-III, is South Korea's next-generation medium-range surface-to-air missile system, succeeding Cheongung-II.
M-SAM Block-III is the third stage of South Korea's medium-range surface-to-air missile family, following the original Cheongung and Cheongung-II (M-SAM Block-II). It is being developed to improve on Cheongung-II's performance against aircraft and ballistic missiles, and is intended to serve in the medium-range layer of the Korea Air and Missile Defense architecture.
AEON carried out UX design and prototyping for the system's operations and engagement control centre.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Integrated air and missile defence is moving away from a world in which each weapon brought its own radar, its own fire control and its own crew. The aim now is a network in which any sensor can cue any suitable shooter, and in which a single command layer sees the whole air picture. The idea is easy to state and difficult to field.
The US Army's Integrated Battle Command System is the furthest along. In an August 2025 flight test, the first live firing using its low-rate initial production hardware, IBCS and the new LTAMDS radar tracked and classified a simulated air-breathing target, computed an engagement and destroyed the target with a PAC-3 MSE interceptor. The same hardware is being deployed to Europe and the Indo-Pacific. Poland got there first among US allies: in December 2025 it declared full operational capability for its IBCS-enabled Wisła system, with two adapted Patriot batteries and plans to expand to eight.
At alliance level, integration is harder. NATO air and missile defence combines systems bought from different suppliers on different timelines, many of them off the shelf. An analysis in European Security & Defence argues that a single integrated network is unrealistic and that integration should instead be organised by threat type, since the demands on latency, range and altitude differ. It notes that Link 16 shares data widely but is not fast enough for close cooperative engagement, and that incompatible battle management software remains a persistent obstacle.
South Korea has followed a national path. Its KAMD Operations Center opened at Osan Air Base in June 2023, replacing the earlier Korea Theater Missile Operations Cell, with the stated aims of cutting the time needed to process ballistic missile data and linking more detection and interception assets. As new long-range and medium-range layers arrive, the ability to allocate threats across them becomes as important as any individual interceptor.
The next step most programmes are pursuing is decision support. With more sensors and shooters on the network, the number of possible engagement pairings grows quickly, and planners are turning to automated aids that recommend which interceptor should take which threat, weighing inventory, probability of kill and the value of what is being defended. Such logic is already embedded in engagement software; the open question is how much authority to give it and how to keep humans meaningfully in control.
For operators, networked IAMD changes the job from running a weapon to managing a shared picture. Crews must understand which sensor a track came from, whether two tracks are the same object, and why the system recommends one shooter over another. The quality of the human-machine interface decides whether automation speeds up decisions or simply moves the bottleneck to the person who has to approve them.

G-Sword (UGV) — Promotional Film

G-Sword is LIG D&A's tracked combat unmanned ground vehicle, unveiled at ADEX 2025.
G-Sword is a tracked combat unmanned ground vehicle (UGV) from LIG D&A, the South Korean defence company formerly known as LIG Nex1, first shown publicly at ADEX 2025 in Seoul. It can be driven remotely from a control console or follow pre-planned routes autonomously, and is offered with modular payloads for roles such as reconnaissance, fire support, close-in air defence and the surveillance of coasts and critical infrastructure.
AEON planned and produced the brand film for LIG D&A's unmanned ground vehicle family, featuring G-Sword.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For years tracked unmanned ground vehicles were a demonstration item at defence shows. The war in Ukraine has turned them into a supply line. By 20 August 2026 Ukrainian units had logged more than 100,000 ground-robot missions since the start of the year, mostly carrying ammunition, food and water forward and bringing wounded soldiers back, and more than 22,000 UGVs had been contracted in the same period. Monthly mission counts rose steadily through the year, a sign that ground robots are becoming routine rather than experimental.
Among established designs, Milrem Robotics' THeMIS remains the reference point for a modular tracked UGV. In October 2025 the Netherlands funded more than 150 THeMIS vehicles for Ukraine, and Milrem reports that 19 countries now use the platform or include it in robotics programmes. Europe is also trying to standardise the layer above the vehicle. The European Defence Fund's iMUGS2 project, launched in Brussels in November 2025 under Estonian leadership, aims at a secure, interoperable architecture for crewed and uncrewed ground platforms, with a possible European unmanned battalion as a long-term goal.
The United States has taken a less direct route. The Army's Robotic Combat Vehicle programme began with light, medium and heavy variants and was narrowed to a single class. In March 2025 Breaking Defense reported that Textron's Ripsaw M3 had been chosen to deliver prototypes, with open questions about autonomy, crew ratios and payloads. Weeks later the Army halted the expected award. The Congressional Research Service reads this as a move towards a cheaper approach involving several vendors, not an end to robotic vehicles, and notes industry criticism of the in-house autonomy software.
The direction is fairly consistent. Bespoke robotic tanks are giving way to common tracked chassis that carry interchangeable payloads: cargo beds, casualty litters, remote weapon stations, anti-tank missiles, sensors, electronic warfare kits or communications relays. Hybrid-electric drives are common, both for quiet running and to power those payloads. Autonomy is arriving in stages, from remote driving to follow-me and waypoint navigation, while open architectures are meant to let armies change software and payloads without changing the vehicle. Ukraine's experience adds a further requirement: the vehicles must be cheap enough to lose.
For operators, the practical questions concern ratios and links. How many soldiers does each vehicle need, and can one operator manage several at once? What happens when jamming cuts the radio link in the middle of a run, and how does the operator know what the vehicle will do next? Control stations that show autonomy state clearly, let payload and mobility tasks be split between people, and keep working in degraded conditions will decide how far these vehicles can be pushed into the most dangerous ground.

Conference — ASW Presentation Materials

Anti-submarine warfare (ASW) is the branch of naval warfare that deals with detecting, tracking and, if necessary, attacking submarines.
Anti-submarine warfare draws on surface ships with hull-mounted and towed sonars, submarines, maritime patrol aircraft, helicopters and sonobuoys, and increasingly on unmanned surface and underwater vehicles. Because submarines are built to avoid detection, the mission depends on combining many sensors over large sea areas and sustaining coverage over time. The Republic of Korea Navy fields destroyers, frigates, submarines, maritime patrol aircraft and helicopters for the mission, and Korean industry develops ASW weapons and sensors.
AEON produced a multimedia conference presentation on how anti-submarine warfare is changing, how it would be operated, and its export prospects.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Anti-submarine warfare is going through its biggest change since the end of the Cold War. Submarines are becoming quieter and more numerous in the Indo-Pacific, and they are being joined by unmanned underwater vehicles. Navies are responding by shifting from a few high-end hunters towards wide networks of sensors, many of them uncrewed, tied together by data processing that would have been impossible a generation ago.
The airborne layer shows the direction clearly. Multistatic active sonar separates the source of a sonar pulse from its receivers: one buoy transmits and a field of others listens for echoes from different angles, which makes a quiet target harder to hide. In April 2026 the US Navy ordered low-rate production of Ultra Maritime's AN/SSQ-125B, a coherent source buoy for the P-8A whose pings are recorded by other buoys in a multistatic field. South Korea sits inside this ecosystem: its navy took delivery of six P-8A Poseidons in June 2024, joining the same family of aircraft, sonobuoys and processing used by the United States, Australia, the United Kingdom and other operators.
Unmanned systems are moving into the barrier role. The UK's Project CABOT envisages commercially operated uncrewed platforms gathering acoustic data in the North Atlantic, with Royal Navy personnel analysing it using AI- and machine-learning-assisted tools, before a second phase brings government-owned uncrewed surface vessels and extra-large underwater vehicles under direct naval control. Similar thinking runs through Australian and US programmes for large autonomous undersea vehicles. The common logic is persistence: drones can stay on station for weeks with a fraction of the crew burden of a frigate.
Exports are the third trend. Korean ASW weapons have already found buyers in Southeast Asia; the Philippine Navy chose to arm its AW159 helicopters with the Blue Shark lightweight torpedo jointly developed by ADD and LIG Nex1. As Korean shipbuilders compete for frigate and submarine programmes abroad, ASW sensors, weapons and combat systems increasingly travel as part of a package rather than as standalone items, and buyers want to know how the whole system will be operated, trained for and supported, not only what each component can do.
For operators, the change is from listening to managing. An acoustic analyst once worked a single array; now the picture is fused from buoys, towed arrays, hull sonars, aircraft and drones, with algorithms doing the first pass. The human role shifts towards deciding which detections deserve attention, allocating scarce assets and explaining a classification to a commander under time pressure. Interfaces that show where each detection came from and how uncertain it is, and that let crews from different services and nations share one picture without confusion, will matter as much as the sensors themselves.

UMEX2026 Vision60 Promotional Film

Vision 60 is a quadrupedal unmanned ground vehicle made by Ghost Robotics, and UMEX is the unmanned systems exhibition held in Abu Dhabi.
Vision 60 (also written Vision60) is a mid-sized, all-weather quadrupedal unmanned ground vehicle developed by Ghost Robotics, a US company majority-owned by LIG D&A. It can be teleoperated or run semi-autonomously, and armed forces have used it for tasks such as base perimeter patrol. UMEX is a biennial unmanned systems exhibition and conference held at ADNEC in Abu Dhabi, and its 2026 edition took place in January 2026.
AEON produced a promotional film on Vision 60 for LIG D&A's presence at UMEX 2026.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The Gulf has become one of the most active markets for unmanned systems, and Abu Dhabi's UMEX is where that is most visible. The 2026 edition, held at ADNEC from 20 to 22 January, was the largest yet, with exhibitors from around 40 countries and a record attendance. It was organised with the UAE Ministry of Defence and the Tawazun Council, with EDGE Group as a strategic partner, and it added live demonstrations, an autonomous drone racing championship and a zone for civilian autonomy.
The UAE is not only hosting the show; it is buying and building. In January 2024 EDGE announced that its Estonian subsidiary Milrem Robotics would supply the UAE Ministry of Defence with 20 tracked robotic combat vehicles and 40 THeMIS ground robots, presented as the largest combat UGV order to date. The package was structured as an experimentation and trial programme, with Milrem training UAE personnel to operate the vehicles, rather than a straight purchase. Saudi Arabia is pursuing a similar logic through its General Authority for Military Industries, which ties drone programmes such as the SkyGuard UAV contract with SAMI to a national goal of localising more than half of defence spending by 2030. Foreign suppliers in the region are increasingly expected to bring local production, training and technology transfer with their products.
Legged robots fit a particular part of this demand. Gulf states guard long perimeters around airbases, ports, energy plants and other critical sites, often in heat and terrain that are hard on people and wheeled vehicles alike. Quadrupeds can climb stairs, cross rough ground and patrol fixed routes while sending sensor feeds back to an operations centre. Explosive ordnance disposal is a second use. In December 2025 Ghost Robotics released a six-axis manipulator arm for the Vision 60 aimed at EOD, door opening and object retrieval, which the robot's software treats as a fifth leg; customers also used it to look around corners at eye level.
The broader direction is towards robots as configurable carriers rather than finished products. Buyers in the Gulf want platforms that can take local sensors, radios and payloads, that can be maintained in-country, and that fit into national command networks. Exhibitions such as UMEX have become venues not just for showing hardware but for signing the partnerships that make that possible.
For operators, security and EOD work both depend on attention over long periods. A patrol robot may run for hours with nothing to report, so the interface needs to bring forward the rare event that matters without tiring the person watching. Remote manipulation is different: handling a suspect object calls for good depth perception, force feedback and camera views the operator can trust. Training, local language support and interfaces that work for crews with different levels of experience will shape how widely these systems are used.
LAMD — Operations Panel 3D Terrain Map UI/UX Consulting
3D UX & Film
LAMD — Operations Panel 3D Terrain Map UI/UX Consulting

LAMD is Korea's low-altitude missile defence system, designed to intercept long-range artillery.
LAMD (Low-Altitude Missile Defence) is a weapon system designed to track and monitor dozens to hundreds of incoming artillery rounds in real time and to engage them within a short time. It is being developed to protect the capital area and major facilities against long-range artillery.
AEON provided UI/UX consulting for the three-dimensional terrain map on the system's operations panel, and produced a film presenting it.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Rockets, artillery shells and mortar bombs are among the oldest threats on the battlefield and among the hardest to defend against. Flight times are short, volumes are high and each round is cheap, so the defender's challenge is less about any single intercept than about sorting a large number of tracks quickly and spending interceptors only where they matter.
The US Army's C-RAM system was an early answer, assembled rapidly for Iraq. It combined existing artillery and air defence sensors, a commercial warning system and a land-based version of the Navy's Phalanx gun, and protected bases in Baghdad and Balad; Northrop Grumman said in 2012 that it had given advance warning of more than two thousand attacks. That figure captures an often overlooked point: in counter-artillery defence, timely warning to the people under the trajectory can save as many lives as the intercept itself.
Israel's layered system carries the idea further. Iron Dome is known above all for selective engagement: its battle management estimates where each rocket will land and engages only those heading for defended areas. In December 2025 Israel's Ministry of Defense and Rafael delivered the first operational Iron Beam high-power laser, which had been tested against rockets, mortars and drones. It is meant to complement Iron Dome, David's Sling and Arrow, adding an option with a very low cost per shot at the bottom of the stack.
South Korea faces this threat at a different scale, with large numbers of artillery pieces and multiple rocket launchers within range of the capital region, where about half the population lives. In April 2026 the government said it would bring forward the LAMD prototype to 2029, two years earlier than planned. The system consists of a radar, a tactical communications element, launchers and interceptors, and is designed to counter simultaneous low-altitude attacks.
Across these programmes the display problem is converging. A counter-artillery engagement can involve dozens or hundreds of simultaneous tracks, each with a predicted impact point, a time to impact and a threat priority. The trend is towards situational displays that compute and prioritise automatically, show predicted impact areas against what is being defended, and let commanders see at a glance which rounds will be engaged and which will be allowed to fall.
For operators, the time available is measured in seconds, so the interface has to do much of the sorting. Crews need to trust the system's prioritisation while still being able to see what it is ignoring and why. Clear visual hierarchy, predictable alerting and a picture that stays readable as the number of tracks climbs are what make a high-volume defence workable for the people running it.

L-SAM-Ⅱ Operations & Engagement Control Centre UX Consulting

L-SAM-II is South Korea's follow-on long-range air and missile defence programme, intended to intercept threats at higher altitudes and longer ranges than L-SAM.
L-SAM-II (Long-range Surface-to-Air Missile II) is a South Korean development programme, led by the Agency for Defense Development, to extend the reach of the L-SAM system. It aims to intercept ballistic missiles at higher altitudes and longer ranges, and includes work on a glide-phase interceptor intended to counter hypersonic weapons. The Defense Acquisition Program Administration formally launched development in January 2025.
AEON provided UX consulting for the system's operations and engagement control centre.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Upper-tier missile defence was built around ballistic trajectories: predictable arcs that can be tracked early and intercepted either in space or on the way down. Hypersonic glide vehicles break that model. After boost they manoeuvre in the upper atmosphere, too low for exo-atmospheric interceptors and too high and fast for most terminal systems, and their final target can remain uncertain until late in flight. The response, in several countries at once, is a new class of interceptor aimed at the glide phase itself.
The most prominent is the US Glide Phase Interceptor. Northrop Grumman was selected as prime contractor in September 2024, and in April 2026 the Missile Defense Agency expanded its contract to accelerate design work. Japan is a partner under a May 2024 agreement and is responsible for rocket motors and propulsion components. The interceptor is meant to launch from Aegis destroyers through the standard vertical launch system, with initial operational capability expected around the end of the decade.
Israel took an earlier step with Arrow 4, announced in February 2021 as a joint programme of the Israel Missile Defense Organization and the US Missile Defense Agency. Its winglets are intended to give it manoeuvrability inside the atmosphere, with hypersonic threats specifically in mind, and it is expected to replace Arrow 2 over time while Arrow 3 continues to cover the exo-atmospheric layer.
South Korea is pursuing the same capability within its own architecture. DAPA formally began L-SAM II development in January 2025, under the Agency for Defense Development and with around nineteen companies involved. The programme aims to intercept at higher altitudes and longer ranges than L-SAM, and its scope includes research on a glide-stage interceptor against hypersonic weapons and intercontinental ballistic missiles.
The interceptor is only half of the problem. A glide vehicle that manoeuvres for long periods must be tracked continuously across sensor boundaries, which is why space-based tracking layers and long-range ground radars are being developed alongside the missiles. Fire control also has to work with predictions that keep changing: the impact point, the intercept window and even the threatened area may all shift during an engagement. These programmes are as much about sensor fusion and battle management as about propulsion.
For operators, glide-phase defence compresses decisions and makes them less certain. Instead of a single predicted impact point, crews may see an evolving envelope of possibilities, and they will often need to coordinate with other layers, or with other services that own the sensors. Interfaces for this kind of engagement need to show uncertainty honestly, make handover between sensors and shooters explicit, and help crews recognise when a prediction has changed enough to warrant a new decision.

ADEX 2025 — Promotional Film

ADEX, the Seoul International Aerospace & Defense Exhibition, is South Korea's largest aerospace and defence exhibition, held every two years.
ADEX brings together Korean and international aerospace and defence companies, government delegations and military visitors for exhibitions, flight displays and business meetings. The 2025 edition was held in October at KINTEX in Goyang, Gyeonggi Province, with around 600 companies from 35 countries taking part.
AEON produced the film shown on the main booth wall of LIG D&A (formerly LIG Nex1) at ADEX 2025, presenting the company's view of the future battlefield with imagery that was generated rather than filmed.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Defence exhibitions have become the main stage on which companies and governments describe how future wars will be fought. ADEX 2025 at KINTEX drew around 600 firms from 35 countries, and its dominant theme was AI-enabled unmanned systems, which coverage tied to Korea's ambition to become the world's fourth-largest defence exporter.
The individual stories varied. Hanwha showed the Themis-K unmanned ground vehicle and talked of turning the K9 howitzer into a combined manned-unmanned system; KAI demonstrated a mixed-reality battle environment with an AI assistant; and exhibitors presented automated air-defence networks and manned-unmanned teaming platforms alongside very high-resolution radar satellites. Across all of them, the future battlefield was presented as a network: sensors in orbit, drones in the air, interceptors on the ground, all joined by software.
The same narrative dominated London a month earlier. At DSEI 2025, NATO's Deputy Supreme Allied Commander Europe argued that mass production and constant innovation are central lessons of the war in Ukraine, pointing to Ukraine's drone output rising into the millions per year. Exhibitors are now expected to show not only individual products but a credible picture of how those products fit into a war of attrition, pervasive electronic warfare and cheap autonomy.
That expectation is changing how exhibition content is made. Future concepts cannot, by definition, be filmed, and many current systems cannot be shown in detail for security reasons. Generative imagery has emerged as a practical way to depict systems and scenarios that do not yet exist or cannot be photographed. It brings its own discipline: defence audiences are expert, and imagery that gets equipment, uniforms or physics wrong loses credibility quickly. The craft is shifting from capturing footage to directing and checking it, with subject-matter review becoming part of the production process.
For operators, the future-battlefield narrative matters because it sets expectations. The common thread across these exhibits is a person supervising many machines through a shared picture, with AI proposing actions. Whether that vision is workable depends on questions rarely visible on a show floor: how many systems one person can realistically oversee, how trust in automation is built and calibrated, and how interfaces present uncertainty when sensors disagree. Companies that can answer those questions convincingly will have a stronger story than those that simply add more drones to the picture.

Future Battlefield UX/CX Consulting (Integrated Unmanned Systems)

Future Battlefield UX/CX Consulting was a study of the near-term battlefield as unmanned systems become part of how forces are composed.
Armed forces are reorganising around unmanned systems as aerial, ground and maritime drones become a routine part of combat. Studies of the near-term future battlefield examine how these systems will be integrated into units, how they will operate alongside crewed forces, and what demand they create for industry.
AEON provided UX/CX consulting to LIG D&A on the integrated unmanned systems battlefield, surveying demand and directions worth developing.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The war in Ukraine has done more than any doctrine paper to change how armies think about unmanned systems. In June 2024 Ukraine made its Unmanned Systems Forces a separate branch of the armed forces, bringing together drone units that had been spread across the army, National Guard and security service. A year later its commander-in-chief said the branch had been responsible for one in every three enemy targets hit in June 2025, and it was being merged with the Drone Line project that coordinates Ukraine's strongest drone units along the front.
The organisational change has been matched by a change in buying. A July 2025 CSIS study found that Ukraine had moved from centralised multi-year procurement to a system in which some 700 front-line units buy directly from approved commercial suppliers, supported by a separate budget for unmanned systems and the DOT-Chain digital marketplace. Private firms take products to a fairly mature stage at their own expense before the military buys them.
The United States has been trying to learn from this. The Replicator initiative, announced in August 2023, aimed to field thousands of low-cost autonomous systems within two years; according to the Congressional Research Service, only hundreds had been fielded by the deadline. Policy then moved further. A July 2025 memo from the Defense Secretary reclassified the smallest drones as consumables, to be handled more like ammunition than aircraft, delegated purchasing authority down to colonels and set a goal of a small expendable drone in every squad by the end of fiscal 2026. In November 2025 the Army said it wanted to buy one million drones over two to three years, against roughly 50,000 a year at the time.
Several lines of force design come out of this. Drones are being treated as a munition rather than a platform, bought in volume and expected to be lost. Units at every echelon are getting their own unmanned capability, while specialist formations take on the most demanding missions. Air, ground and maritime systems are expected to work together, and counter-drone defence has become a requirement for every unit rather than a specialist task. Industry is being asked to deliver at commercial speed, with software and payloads that can be updated in the field.
The constraint that shows up everywhere is people. Every drone, however cheap, needs someone to launch, guide, supervise or recover it, and trained operators are scarce. That puts pressure on interfaces that non-specialists can learn quickly, on control stations that let one person manage several vehicles, and on ways to deconflict crowded airspace full of friendly drones. As autonomy takes on more of the flying, the human role moves towards judgement: choosing targets, setting limits and deciding when to intervene. Designing for that role is now part of designing the force.

M-SAM Block-II — Control Centre UI Localisation (UAE)

Cheongung-II (M-SAM Block-II) is a South Korean medium-range surface-to-air missile system, exported to the United Arab Emirates, Saudi Arabia and Iraq.
Cheongung-II, also designated M-SAM Block-II, is a medium-range surface-to-air missile system developed in South Korea to intercept aircraft and ballistic missiles. The United Arab Emirates ordered the system in 2022, and Saudi Arabia and Iraq followed with orders in 2024. In the UAE it operates within a wider air defence network that also includes Patriot.
AEON localised the user interface of the system's control centre for the UAE version.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
South Korea has become one of the more significant new suppliers of ground-based air defence, and the Gulf is where that shift is most visible. Medium-range systems sit at the core of this market, because they cover aircraft, cruise missiles and shorter-range ballistic missiles, and because they must fit into networks that already include American upper-tier and lower-tier systems.
Cheongung-II is the clearest example. The UAE ordered ten batteries in 2022 and integrated two into an air defence network that also includes Patriot. In early 2026, during Iranian strikes across the region, the system saw its first combat use abroad; the UAE then asked for more than thirty additional interceptors, and Seoul brought the delivery forward. The same report notes that other Gulf states, seeking faster deliveries, have reportedly approached Italy about SAMP/T.
Orders followed elsewhere in the region. Iraq's contract in September 2024 made it the third export customer, after the UAE and Saudi Arabia. As of mid-2026 Saudi Arabia has ten batteries on order and Iraq eight, with deliveries beginning in 2028, and interest has spread beyond the Middle East, with a letter of intent from Indonesia and reported Swiss interest. The Korea JoongAng Daily points out that the consortium builds about eight batteries a year, so production capacity rather than demand is now the limiting factor.
Exported systems rarely operate alone. Gulf buyers already field American equipment: Lockheed Martin lists THAAD batteries in service with both the UAE and Saudi Arabia, alongside the region's long-established Patriot fleets. A new medium-range layer therefore has to exchange tracks and coordinate engagements with equipment from other suppliers, through common data links and the buyer's own national command structure. In practice, interoperability is part of what is being sold.
Localisation sits alongside interoperability. A system designed around one army's doctrine, language and procedures has to be adapted to another's, often with right-to-left script, different terminology and different rules for engagement authority. This goes well beyond translating labels: it touches training material, alert wording, units of measure and the way the system expresses priorities and warnings. Buyers increasingly expect this adaptation, together with local support, as a condition of purchase.
For operators, the result is a working environment assembled from several national traditions. A crew may work with consoles built by different companies in a single shift. Interfaces that respect the operator's language and doctrine reduce cognitive load and error, especially under stress, and consistency in alerts and track symbology across systems matters as much as the capability of any individual console.

Samsung Cloud Web UX

Samsung Cloud had been sharing its ground with someone else's drive. Standing on its own meant deciding what it was for.
Separating from OneDrive raised a question backup itself never has to answer: what does a person come to the web for, when their data lives on a phone?
Mostly the gallery. So the site was built around photographs and what happens to them — syncing, backing up, and the subscription that pays for the space.
And one account is rarely one device. A current phone, an older one still in a drawer, a tablet, a watch, each with its own idea of what has already been saved. The structure has to hold all of them without asking the person to keep track.
Research into how cloud services present themselves on the web, several design proposals, and a responsive HTML guide — all on the One UI base.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For most people the cloud stopped being an abstract place for files some time ago. What they actually keep there is a camera roll, and the services that hold those libraries have spent the past five years renegotiating two things at once: what storage is worth, and what a library is supposed to do once it is far too large to scroll.
A turning point for many users came when Google ended free unlimited uploads to Google Photos, announced in November 2020 and effective from June 2021, after which new photos counted against the storage shared with Gmail and Drive. The change made something visible that had been hidden for years: a growing photo library is a cost as well as an archive. Quotas, duplicates and clean-up tools moved from the edges of these products towards the centre.
The second shift is in how libraries are organised. Apple described its iOS 18 update as the biggest redesign of Photos to date, replacing a set of tabs with a single grid and collections the system assembles by itself, such as trips and recent days. Google went further with Ask Photos, a Gemini-based feature introduced in May 2024 that lets people question their library in everyday language. The album, a container the user built by hand, is giving way to structures the service infers on the user's behalf.
Regulation is pulling in a different direction, towards mobility. The EU Data Act, applicable since September 2025, sets out a framework for customers to move between data processing services, and export tools such as Google Takeout have become a routine part of account settings. A personal archive that was once tied to whichever brand of phone a person happened to buy is increasingly expected to travel with them.
The web version of these services has quietly changed role as well. Capture and backup now happen automatically on the phone, so the browser is less a daily destination than a place for the jobs that need a large screen and a keyboard: reviewing thousands of items at once, deleting in bulk, downloading an archive, or sharing with someone outside the ecosystem.
For users, the result is a library that organises itself more each year and asks to be understood less as a set of folders than as a service with its own judgement. Interfaces in this space serve people well when they make that judgement legible: what has been saved and where, why an item appears in a collection, and how to take everything out again. Storage people cannot see into is storage they gradually stop trusting.

Frigate, OPV Combat System UX Consulting (Peru)

The Peruvian Navy is acquiring a frigate and an offshore patrol vessel designed by South Korea's HD Hyundai Heavy Industries and built in Peru by the state shipyard SIMA.
In April 2024 HD Hyundai Heavy Industries signed a contract with SIMA, Peru's state shipyard, for a frigate, an offshore patrol vessel and two landing craft for the Peruvian Navy. HD HHI supplies the designs, equipment and technical support, and SIMA builds the ships locally. The frigate is derived from HD HHI's HDF-3200 design and the patrol vessel from its HDP-2200 design.
AEON provided UX consulting on the combat system interface for the frigate and the offshore patrol vessel.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
South Korea's naval shipbuilders have spent the past decade turning domestic frigate and patrol-vessel designs into export products, and Latin America is the newest market. In April 2024 HD Hyundai Heavy Industries signed with Peru's state shipyard SIMA for a frigate, an offshore patrol vessel and two landing craft, supplying the designs, equipment packages and technical support while SIMA builds the hulls locally. The agreement also made HD HHI a long-term strategic partner in Peru's naval force development, and the Peruvian Navy has signalled that more frigates and patrol vessels may follow the first batch.
The model was refined in Southeast Asia. The Philippine Navy's two Jose Rizal-class frigates came from HD HHI, followed by a run of offshore patrol vessels and, in December 2025, a contract for two further 3,200-tonne frigates based on the same HDF-3200 design family, due by 2029. Repeat orders of a proven hull appeal to medium-sized navies: schedules are shorter, training carries over, and spares and maintenance can be shared across a class. Thailand took a similar path earlier with a frigate from DSME, now Hanwha Ocean.
The more revealing part of these programmes is the combat system, because that is where an exported hull becomes a warship, and Korean exports have taken different routes. Thailand's HTMS Bhumibol Adulyadej was built in Korea but fitted with Saab's 9LV combat management system. The Philippine ships use Hanwha Systems' Naval Shield, and Janes reports that the newer Philippine patrol vessels carry a later Naval Shield baseline than the frigates, with lighter consoles. For Peru, LIG Nex1, now LIG Defense & Aerospace, is supplying the combat systems, electronic warfare equipment and data links for the frigate and patrol vessel, and in July 2026 the company opened a regional office in Lima to support customers across Latin America.
The direction of travel is clear. Korean firms increasingly compete on the whole package rather than the hull alone: platform, combat management, sensors, data links, weapons and through-life support, usually with local construction or industrial participation as a condition of sale. Buyers, for their part, want sovereign access to software, interoperability with partner navies and data-link standards, and an upgrade path that does not leave them tied to a single foreign supplier. Modular combat-system architectures and in-country support offices are the industry's response, and they make a combat system a long-term relationship rather than a one-off delivery.
For operators, an imported combat system is where a navy's existing doctrine, training and language meet a foreign design. Localisation goes well beyond translating labels: alerting philosophy, task sequences and terminology have to match how that navy trains and fights, and they have to stay consistent across a class that may be built in more than one yard over a decade or more. The quality of that fit shapes how quickly crews reach confidence with a new ship.

UAV Ground Control System UI/UX Development

A UAV ground control station multifunction console is the operator station from which several unmanned aerial vehicles are commanded and monitored.
A ground control station (GCS) is the land-based facility from which unmanned aerial vehicles are commanded and through which their flight data and sensor feeds are received. A multifunction console lets operators handle vehicle control, payload operation and mission monitoring from one workstation, and newer stations are designed to manage more than one aircraft at a time. Within NATO, the interfaces between control stations and air vehicles are standardised under STANAG 4586.
AEON developed the UI/UX for LIG D&A's ground control station multifunction console for operating multiple UAVs.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For most of the drone era, one aircraft meant one crew and one control station, usually supplied by the aircraft's maker. NATO tried to loosen that link early. STANAG 4586 sets out standard interfaces between UAV control systems, air vehicles and wider command systems, so that a station built by one nation or company can work with another's aircraft at agreed levels of interoperability. The standard itself is careful to say that compliance enables interoperability but does not guarantee it; operator proficiency and concepts of operation still decide what works in practice.
The bigger change now is the ratio of people to aircraft. In December 2023 US Air Force Special Operations Command and General Atomics showed a single crew controlling three MQ-9A aircraft from one workstation using a government-owned control suite. The aircraft took off under separate ground stations, were handed to the single workstation once airborne, and were passed back for automatic landings. AFSOC presented it as a way to reduce manpower.
The Air Force is applying the same idea to its Collaborative Combat Aircraft. Its Autonomy Government Reference Architecture is a government-owned interface between mission autonomy software and the aircraft. In February 2026 the service reported that Collins Aerospace software had flown a semi-autonomous mission on General Atomics' YFQ-42A for more than four hours, following commands from a ground operator, with Shield AI software due to fly on Anduril's YFQ-44A. The stated aim is to avoid being locked into one vendor's software. The Army is moving the control node into the air as well: at Project Convergence Capstone 6 in July 2026 an H-60M Black Hawk launched and controlled several air-launched effects at the same time.
Ukraine shows the same need at the low end. Front-line operators there often carry several different controllers because each drone maker supplies its own. In April 2026 the Ministry of Defence announced a universal ground control station for fibre-optic drones, built to a common standard agreed with manufacturers, to replace them.
The pattern across these efforts is consistent: the control station is separating from the aircraft. Governments want to own the interfaces, run several vehicle types from one console, and change autonomy software without rebuilding the station. That favours open architectures, common message sets and modular software.
For operators, the role changes from flying to supervising. A person responsible for several aircraft cannot hold every detail at once, so the console has to manage attention: showing the state of each vehicle at a glance, raising alerts in order of urgency and making handovers between stations explicit. When different aircraft and autonomy packages share one console, a consistent interaction model becomes a safety issue, because an operator under pressure should not have to remember which vehicle behaves which way.

AI Powered Website Navigation Prototype

A design site with several hundred pages does not get searched. It gets given up on.
design.samsung.com held more than anyone could find in it. The question was not how to index it better but whether a person could simply ask.
Three decisions carried the prototype. Whether answers may come only from the site's own content or from anywhere. How a source is shown, so that an answer can be checked rather than believed. And how little screen a conversation actually needs.
We copied the whole site, connected it through retrieval, and built a plain conversational surface. Most of the effort went into whether the answers were right rather than into how the screen looked — a prototype that demonstrates well and answers badly teaches the wrong lesson.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The idea of asking a website a question instead of navigating it is older than the current wave of AI, but it only became practical once language models could be tied to a specific body of text. Most deployments descend from retrieval-augmented generation, described by Lewis and colleagues in 2020: a retriever pulls relevant passages from an index and a generator writes a response from them. Six years later the pattern sits behind help centres, documentation portals and internal knowledge tools alike.
The wider search landscape has shifted around it. Generated summaries now appear above conventional results on major search engines, and the effect on behaviour is measurable. A Pew Research Center analysis of browsing data from March 2025 found that people clicked a conventional result on 8% of visits to Google pages showing an AI summary, against 15% on pages without one, and followed a link inside the summary on only 1% of visits. Fewer visitors reach a given site at all, and those who do arrive with expectations set by answer engines.
On-site assistants, however, have struggled to find their place. Nielsen Norman Group research published in 2026 found that users rarely see a reason to open a site's chatbot. Entry points are easy to miss, earlier bots taught people to expect scripted loops and deflection, and search, filters and menus are often faster for comparing options. The assistants that earned their keep handled quick, specific questions, or drew together information that would otherwise mean reading several pages.
Trust is the harder problem. In a large preregistered experiment, Li and Aral found that reference links increase trust in generated answers even when those links are wrong or invented. A citation therefore works as a signal of credibility whether or not anyone follows it. That shifts responsibility onto the system rather than the reader: retrieval quality, regular evaluation against known questions and a candid response when the material simply does not contain an answer.
Organisations are also learning that a conversational layer exposes the state of the content underneath it. Outdated pages, contradictory guidance and gaps that a menu could hide become visible the moment someone asks a direct question, which is why many such projects turn into content audits as much as interface work.
For people using these tools, the change is from finding to judging. Navigation asked them to learn how a site was arranged; conversation asks them to assess an answer they did not watch being assembled. Interfaces that serve users well make an assistant's scope clear before the first question, keep ordinary search and navigation within easy reach, and help people tell a grounded answer from a confident guess.

Content Design Research

A notice is written once and arrives on a watch, a refrigerator and a television. The words do not change. Everything around them does.
Samsung makes more kinds of device than any single guideline has had to cover. The text that addresses the user — a system update, a notice, a caution, a warning — appears on all of them, and it is the part people actually read.
One UI already set the base, so the difficulty was never the rule. It was the range. Three things decide how a message lands: where it sits, when it appears, and the tone it takes — and all three shift with the screen. A sentence that reads as calm on a phone reads as an emergency on a watch face, and goes unnoticed on a television across the room.
We gathered how Apple and Google introduce new features across their television, mobile and watch systems, and read the examples for exactly those three. Features arrive faster than they used to, so the moment of introduction is now a large part of whether a function is ever used.
Consistency has a second axis besides the device. What counts as sufficient warning in one market reads as alarming in another, or as evasive. That is a habit, not a translation problem, and it decides whether the same caution is actually heard.
A research report, the writing itself, the design guide, and a prototype.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Content design has moved from the margins of product teams to a recognised discipline over the past decade. The UK Government Digital Service did much to define it, framing the work as starting from what people need to find out or do and only then deciding format, length and placement. Large technology companies have since built content design teams of their own, and the remit now covers system text such as alerts, permission requests and onboarding as much as product copy.
Platforms have meanwhile taken more control over whether that text is seen at all. Apple's iOS 15 introduced Focus modes and a scheduled notification summary that hold back non-urgent alerts and rank the rest using on-device signals. Android 13 went a step further by making notifications a runtime permission for newly installed apps, which must now ask before sending most alerts. A message no longer reaches someone simply because an app chose to send it; it has to pass the operating system's filters and the person's explicit consent.
Accessibility expectations add another layer. The W3C's guidance on making content usable for people with cognitive and learning disabilities asks for familiar words, short sentences, literal language and clearly separated instructions, principles that help far more people than the group the document was written for. In Europe, the European Accessibility Act covers smartphones, computers and their operating systems, television equipment and e-books, and its requirements have applied since June 2025, bringing the words on consumer devices within the reach of law as well as style guides.
The newest pressure comes from generative AI. Operating systems now summarise, condense and reorder text that product teams wrote, and early deployments showed how easily a shortened message can lose or distort its meaning. That makes the source text more important, not less: a notice that is specific and unambiguous survives machine summarisation better than one that leans on context or tone to be understood.
Taken together, these shifts change what a content designer is accountable for. The job is no longer only to write a good sentence, but to anticipate the filters, permissions, assistive technologies and models that will stand between that sentence and the person meant to read it.
For readers, the trend cuts both ways. Fewer interruptions reach them, but those that do carry more weight and arrive already sorted by systems they have not configured in detail. Good system text respects that position: it says what happened and what, if anything, the reader needs to do, in words that still make sense when glanced at, read out of context, or shortened by a machine.

DA SmartThings UI Guidelines

The same function now lives in two places — the app in your hand and the panel on the machine. When they disagree, the person is the one who has to reconcile them.
SmartThings reached the point where an appliance could be worked from a phone or from its own screen, and both had to be right.
The decision is not what to unify but what is allowed to differ. A phone can scroll and a washing machine cannot; pretending otherwise makes both worse. So: what matches exactly, and where a device's own nature may show through.
It built on guideline work carried forward from earlier years, widened this time to take in the set screens as well.
4 separate engagements on this line, 2021–2024.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In the mid-2020s the largest platform owners each moved towards a single design language spanning all their screens. Apple's June 2025 redesign was described as the first to extend across iOS, iPadOS, macOS, watchOS and tvOS while keeping what is distinct about each. Samsung took a parallel step for the kitchen and laundry room two months later, extending its One UI platform to home appliances so that phones, TVs and appliances share common interface elements, apps and services, with up to seven years of software updates for Wi-Fi models launched from 2024. The appliance is no longer treated as a separate island of interface design. It is one more surface in a family.
The case for this rests on an old usability principle. Nielsen Norman Group distinguishes internal consistency within a product family from external consistency with conventions people meet elsewhere, and argues that both reduce the effort of learning and give people confidence that the same element will behave the same way. A household that owns a phone, a television and a refrigerator from one maker reasonably expects an icon, a colour or a gesture to mean the same thing on each.
A shared language does not mean identical screens, however, and the Apple announcement was careful to speak of harmony rather than sameness. An appliance panel is fixed in place, often small, used with wet or full hands and sometimes read from across a room, while a television is used from a distance with a remote and a watch is used in seconds. What platform owners tend to share across such different surfaces is the foundation: colour, type, iconography, motion and the names of things. Layout and interaction stay specific to each class of device, and design systems increasingly document those platform-specific rules alongside the shared ones.
The tooling has caught up with the ambition. In October 2025 the W3C Design Tokens Community Group released the first stable version of its specification, a vendor-neutral format for design decisions with support for theming and multiple brands and with output to several platforms. It is a community group report rather than a formal W3C standard, but several major design tools already support it. For a manufacturer maintaining dozens of product screens alongside a phone app, tokens mean that a decision about colour or type can travel from one source to every surface, rather than being copied by hand and drifting apart.
For the people living with these products, a unified design language pays off mainly in recognition: an icon learnt on the phone is understood on the television, and a settings structure met once does not have to be learnt again. The risk is a surface-level consistency, where screens share a visual style but differ in how they behave, which can mislead more than an honest difference would. A useful measure of the trend is therefore not how alike the screens look, but how reliably people can predict what will happen when they act on any of them.

for AI UX Scenario Film Development

Every advance costs somebody something. The film had to show who, and what an assistant could take off them.
The word is technostress, and it does not fall evenly. It lands hardest on people the ordinary path already excludes — a deaf user, an older user, anyone for whom the obvious way in does not open.
The argument lives or dies on which people and which moments you choose to show. Pick the wrong scene and inclusive AI reads as a feature list; pick the right one and it needs no explaining.
So we wrote scenarios in which a multimodal assistant — one that can see the screen, hear the room and act on the device — takes the difficulty off the person instead of adding to it.
Then we built what the film needed in order to be true rather than illustrated: the key screens, the assistant's own graphics, the micro-interaction motion, and the composite that puts all of it back into the footage with sound.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Assistive technology has always been a larger field than the products marketed under that name. The WHO and UNICEF's global report on assistive technology, published in May 2022, estimated that more than 2.5 billion people need at least one assistive product, a figure expected to reach 3.5 billion by 2050 as populations age. The report counts software and apps that support communication and cognition alongside wheelchairs and hearing aids. Since then, general-purpose AI has begun to blur the line between an assistive product and an ordinary phone.
An early sign came from Be My Eyes, a service that connects blind and low-vision people with sighted volunteers over video. In March 2023 it introduced a Virtual Volunteer built on OpenAI's GPT-4, later renamed Be My AI, which lets a user send a photo and ask questions about it, with the option of reaching a human volunteer when the model cannot help. The design choice is telling: the model did not replace the human network, it became the first line in front of it.
Platform makers, meanwhile, built narrower capabilities into the device itself. In May 2024 Apple announced Eye Tracking, which lets people with physical disabilities control an iPhone or iPad with their gaze using the front camera, with setup data kept on the device. It came alongside Listen for Atypical Speech, which adapts recognition to people with conditions such as cerebral palsy, ALS or stroke, and Music Haptics, which renders songs as taps and vibrations for deaf and hard-of-hearing users. These are dependable, bounded functions rather than open-ended assistants.
The next step is assistants that watch and listen continuously. Google DeepMind has developed a Visual Interpreter prototype of Project Astra with the visual interpreting service Aira, describing surroundings as the camera moves and drawing on Maps, Photos and Lens. Notably, it is offered through a trusted-tester programme with live professional human oversight. That caution reflects the open questions: a model that describes a street confidently but wrongly can do more harm than one that stays silent, and an always-on camera raises privacy issues for bystanders as well as users.
For people who rely on these tools, the shift is from learning a device's interface to stating a goal and letting the system find the route. That can remove a great deal of everyday friction, but only if the assistant is honest about uncertainty, easy to interrupt and able to hand over to a person. Designers now have to decide when an assistant should act and when it should ask, and how someone who cannot see or hear the screen can tell what has just been done on their behalf.

Digital Legacy

The person who set this up is gone. The people using it are grieving, and some of them are old.
Digital Legacy hands a person's data to their family after they die. That sentence is the whole design problem: the user is not the customer, and they arrive at the worst moment of their year.
A death has to be evidenced, a temporary account issued to the family, and the data made available to download. Each of those is a form somebody is filling in while grieving. Behind it sits a console showing the applications as they arrive, where each has got to, and the documents attached.
How few steps can a transfer of this kind survive? Security wants proof, and proof means forms; grief has no patience for them. Every screen sits somewhere on that line, and the position has to be chosen rather than defaulted into.
The path runs long before it runs fast — finding the feature, naming an heir, changing your mind, and only then the part that happens afterwards. We began with desk research and a content guide, since here the words come before the layout, and carried it through interface, graphics, a responsive HTML guide and a working prototype. Cloud and account behaviour, and the handover between phone and web, had to hold across all of it.
2 separate engagements on this line in 2024.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
What happens to a person's accounts after they die used to be settled, if at all, through a support ticket and a scanned letter. Over the past decade it has become a product area in its own right, shaped by platform policy, inheritance law and the plain arithmetic of an ageing online population.
The arithmetic is striking. Researchers at the Oxford Internet Institute projected in 2019 that deceased users could outnumber living ones on Facebook within roughly half a century, even under a scenario in which the network gained no new members. Whatever the exact date turns out to be, every large service already holds accounts whose owners have died, and relatives meet those accounts through shared photos, message histories and reminders that keep arriving.
Platforms have converged on advance nomination. Meta lets people choose a memorialised account manager who can look after a profile without being able to log into it, or ask for the account to be deleted instead. Apple's Legacy Contact, introduced with iOS 15.2, gives a nominated person access to photos, messages, notes, files and backups once they present an access key and a death certificate, while passwords, payment details and purchased media stay out of reach. Google's Inactive Account Manager works on a timer, passing chosen data to trusted contacts after a period of silence.
These tools now sit alongside stricter housekeeping. In 2023 Google announced that personal accounts left unused for two years may be deleted, citing the security risk of dormant accounts, with removals starting from December that year. Policies of this kind make planning less optional: an account nobody prepared for may simply disappear. Legislation varies widely between countries, and in many places the terms of service remain the working rulebook.
An unusual property of these features is their timescale. They are set up in a few taps and then left untouched for years, which raises questions that ordinary products rarely face: whether a nomination is still current, whether the nominee even knows about it, and whether an access key survives a lost or replaced phone. Prompting people to review such settings, without turning mortality into a nag, is becoming part of the job.
For the people involved, the interaction is less about speed than about clarity and dignity. Interfaces in this area work best when they state plainly what a nominee will and will not be able to see, use language that assumes no technical confidence, and give everyone a predictable picture of what happens next. Few products are used so rarely and matter so much when they are.

LIG Nex1 Groupware Screen Sharing Service Build Publishing

A groupware system built more than a decade earlier, inside a defence company, for people who are frequently not at their desks.
A renewal inherits the original build. Much of what could be changed had been settled years before by decisions nobody was there to argue with.
In a defence company security is not a setting; it is the condition the work happens under. Moving between a desk machine and a phone has to stay safe without becoming unusable for someone who is travelling — which is most of them, most of the time.
Front-end publishing for the screen-sharing service.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Screen sharing became ordinary almost overnight. What had been a feature of specialist conferencing software through the 2010s turned, during the pandemic, into a daily routine for office workers, and the technology underneath it matured to match. In January 2021 the W3C and IETF declared WebRTC an official standard, formalising the browser-native real-time audio and video that had carried much of the previous year's meetings, teaching and interviews.
Sharing a screen, though, is a different act from sharing a camera. The W3C's Screen Capture specification, which defines how a web page asks to capture a display, a window or a tab, is unusually direct about the risk: anything visible on the captured surface, including confidential material from other sources, can leave with the stream. Its safeguards centre on the person sharing, who must choose the surface each time, consent again for every capture, and be able to see clearly that capture is live and stop it.
Android has moved in the same direction. Starting with Android 14 QPR2 and continuing in Android 15, users can share a single app instead of the whole screen; consent is required for each capture session, a status-bar chip shows any ongoing projection, projection stops automatically when the device locks, and apps can detect that they are being recorded so they can warn before a sensitive step.
Enterprise tools add a further layer aimed at what happens once content reaches the viewer. Microsoft Teams, for example, lets administrators require watermarks on shared content and video that show each participant's own email address, to discourage screenshots. The trade-offs are explicit: some meeting features switch off while a watermark is in use, and participants on unsupported clients are limited to audio. Organisations decide the balance meeting by meeting or through sensitivity labels.
Behind these features is a broader shift in how organisations think about access. NIST's zero trust architecture guidance, SP 800-207, published in 2020, argues against granting trust on the basis of network location and for checking users, devices and requests against the specific resource being protected. Applied to screen sharing, the question is no longer only whether someone is inside the corporate network, but whether this person, on this device, should see this content at this moment.
For the people presenting and watching, the result is a more deliberate interaction. Good screen-sharing interfaces make the shared surface unmistakable, show at a glance who can see it, make stopping instant, and explain restrictions such as watermarks or audio-only fallbacks before they surprise someone in the middle of a meeting. Where rules are strict, clarity about what is being shared is what lets people use the feature with confidence instead of avoiding it.

L-SAM Operations/Engagement Control Centre Film

L-SAM is South Korea's long-range surface-to-air missile system, whose development was completed in 2024.
L-SAM (Long-range Surface-to-Air Missile) is a South Korean system for intercepting ballistic missiles and aircraft at higher altitudes than earlier Korean air defences, developed by the Agency for Defense Development with domestic industry. Development was completed in 2024, after which the programme moved towards production, with deployment expected in the second half of the 2020s.
AEON produced the film reporting the completion of the system's development.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
An air-defence programme does not end when a test missile hits its target. Between a successful intercept and a battery standing watch lies a sequence of milestones, each with its own meaning: development testing, operational testing with service crews, a production decision, initial operational capability and, eventually, full operational capability. The gaps between them are where most of the real work of fielding happens.
The US Army's Integrated Battle Command System shows how long that path can be. Its initial operational test and evaluation ran for ten months at White Sands Missile Range, with soldiers detecting, tracking and intercepting missiles and surrogates while under electronic attack. Full-rate production was approved in April 2023 and initial operational capability followed in May, which, as DefenseScoop reported, cleared the system for fielding to units while further capabilities continued to be added. IOC, in other words, was a licence to start fielding, not a declaration that the system was finished.
Germany's Arrow 3 followed a compressed version of the same logic. Ordered in 2023, it entered service in an initial configuration at Holzdorf/Schönewalde, south of Berlin, in December 2025, with further sites in the north and south of the country planned by the end of the decade. The first site provides initial coverage, but the full architecture of three batteries, and the crews to run them, takes years more.
South Korea's long-range programme is moving through the same stages. The defence ministry declared L-SAM development complete in November 2024, with production expected to begin in 2025 and deployment in the second half of the decade, while a Block-II version for higher-altitude interception is developed in parallel. The interval between development and deployment is where doctrine is written, crews are trained and a new layer is integrated with the existing air and missile defence network.
Two trends are reshaping this process. First, testers are bringing soldiers in earlier, because problems found by crews at the end of development are the most expensive to fix. Second, software-defined systems keep evolving after IOC, so the old idea of a fixed baseline that is tested once and then fielded is giving way to a stream of software releases, each needing its own verification. Programme milestones increasingly mark points in a continuous process rather than finish lines.
For operators, this means the system they train on is rarely the system they will retire. Interfaces, procedures and capabilities shift over the service life, and crews are often the first to discover how a new release behaves under pressure. Designing for this means keeping interaction patterns stable even as functions are added, explaining changes in terms crews can use, and treating operator feedback from exercises and test campaigns as a primary input to each release.

2.25LCD UI Common & Product Guidelines Collaboration

One 2.25-inch panel, many appliances. The screen is the same size everywhere and the machine behind it never is.
A guideline for a display this small is mostly a question of what it is allowed not to say. The work was drawing the line between what every product must share and what each one may keep for itself — written as a common guideline with product guidelines sitting under it.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Walk through a large appliance showroom in 2026 and the displays sort themselves into tiers. At the top are refrigerators with door-sized touchscreens. Below them come washers and ovens with mid-sized panels. Across most of the range sit small fixed displays that show a cycle, a temperature or a countdown, and very little else. Samsung made this tiering explicit when it announced that its AI Home screen would come in 9-inch and 7-inch versions across refrigerators, laundry and wall ovens, with a 4.3-inch laundry variant in some markets whose features are not identical. For interface designers the interesting part is not the large screen but the ladder. The same service now has to be expressed at several sizes, and someone has to decide what survives each step down.
That decision is mostly about how much a person can take in. Nielsen Norman Group argued as early as 2014 that design across screen sizes is limited by the capacity of the human–device channel: smaller screens show less at once, so people hold more in working memory, and their attention is more often interrupted. An appliance makes those limits tighter still. Its display is read in a glance, often while someone is carrying a basket or minding a pan. Small displays therefore lean on a compact vocabulary of icons, numerals and short labels, and international symbol standards and long-established conventions, such as the power symbol, do much of the work of making that vocabulary readable without instruction.
The second shift is that an appliance's software no longer stays as it was on the day of purchase. LG's ThinQ UP programme, launched in Korea in January 2022 and rolled out internationally from 2023, lets owners download new appliance functions through the companion app. Samsung described a wall oven whose plain 7-inch LCD was upgraded over the network into its richer home screen. Once features can arrive after launch, a display guideline can no longer be a set of finished screens. It becomes a set of rules and reserved places that a function nobody has designed yet can fit into.
Running against the spread of touchscreens is a renewed respect for controls that can be found without looking. The clearest signal comes from cars rather than kitchens: Euro NCAP's 2026 protocols reward dedicated physical controls for indicators, hazard lights, wipers, the horn and the emergency call, treating touch-only operation of core functions as a safety cost. Appliances face a gentler version of the same question. Wet hands, a glance from across the room and a hurried morning all favour a control whose position and feedback do not depend on reading a screen.
For the people who use these machines, the trend points to a few modest virtues. A display should state the machine's condition plainly, in a form that can be taken in at a glance. Functions added by an update should turn up where people already look, rather than behind a menu they have never opened. And the most frequent actions, such as start, pause and cancel, should remain possible by feel, whatever the screen above them is able to show.

LIG Nex1 Groupware & Integrated Support Enhancement UX/UI Development

A groupware system more than a decade old, patched on request and never once rebuilt.
For over ten years it absorbed whatever individual departments asked for, one request at a time. The code underneath had become as difficult as the screens above it, and the age of the thing showed everywhere a person touched it.
The worst of it was the phone. A defence company keeps mobile access behind its own security boundary, so the little that could be done from a handset took several authentications to reach — and the people who needed it most were the ones travelling, furthest from a desk.
Integrated support, in the end, meant the mobile side: making a phone a real way into the system rather than a reduced copy of it. The users were every employee and officer in the company, so there was no group small enough to leave behind.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Enterprise sign-in has been substantially rethought in the past few years. For a long time, protecting corporate systems meant accumulating checks: a password, then a one-time code, then a push approval, each added for good reason and each adding friction. Current security guidance treats much of that stack as weaker than it looks, and points instead towards fewer, stronger steps.
The clearest statement came from the US Cybersecurity and Infrastructure Security Agency. In October 2022 CISA described phishing-resistant MFA as the gold standard, ranking FIDO/WebAuthn and PKI-based methods above one-time codes and push prompts. It noted that codes can be phished and that plain push approvals invite push bombing, in which a user is flooded with requests until one is accepted. The same guidance has a practical edge for older estates: it asks organisations to list systems that cannot support modern MFA, connect them to enterprise single sign-on where possible, and otherwise plan their replacement.
NIST followed with its revised Digital Identity Guidelines, SP 800-63-4, finalised in July 2025. The authentication volume makes phishing resistance a defined property of cryptographic authenticators, requires services at its middle assurance level to offer at least one phishing-resistant option, permits syncable authenticators such as passkeys below the highest level, and keeps usability in view: no forced periodic password changes, no composition rules, password managers and pasting allowed, and an explicit warning about authentication fatigue.
The device side has changed too. Rather than treating a phone as either wholly corporate or wholly personal, Android's work profile creates a separate, IT-managed space for company apps and data on the same handset, with its own storage and administrator control over what may cross between the two sides. Combined with zero trust thinking, which evaluates each request rather than trusting the network it arrives from, this lets organisations extend more services to personal and company handsets without taking control of the whole device.
The cumulative effect is that strong security and short paths are no longer opposites. A single phishing-resistant sign-in, backed by a managed profile and device checks that run in the background, can replace several visible steps. What remains hard is organisational: deciding which older services to connect and which to rebuild, phasing roll-outs so that administrators and help desks go first, as CISA suggests, and making sure that a lost or replaced phone does not lock someone out of their working day.
For employees, the change should show up as fewer interruptions and clearer prompts. That depends on interfaces that ask for proof once and remember it sensibly, explain plainly when a further check is needed and why, keep work and personal spaces visibly distinct, and offer recovery that people can complete on their own.

Windows For ifland Renewal (AI including) Planning and Design

A renewal with AI folded into a service people already knew how to use.
Where the AI goes is decided by what it must not disturb. The existing path through the service was the constraint, not the canvas. Planning and design for the PC version.

UI Design Consulting (Naval Acoustic Information Management System (NAIMS))

The Naval Acoustic Information Management System (NAIMS) is a system for managing acoustic information collected by surface ships and submarines.
The Naval Acoustic Information Management System (NAIMS) manages acoustic information collected by surface ships and submarines. Acoustic information of this kind underpins the detection and classification of contacts in anti-submarine and undersea warfare, and a managed store of it supports analysis, training and the upkeep of reference data.
AEON provided UI design consulting for the system.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Acoustic data is the raw material of undersea warfare. Sonar operators classify contacts by comparing what they hear, and what they see on frequency and time displays, with signatures gathered over years of collection, and navies guard those libraries closely. What is changing is volume. Towed arrays, hull sonars, sonobuoy fields and a growing number of uncrewed sensors now produce more acoustic data than human analysts can review, and the response almost everywhere is machine learning.
The most visible programme is the trilateral AUKUS effort. In 2024 the partners described an AI tool for their P-8 maritime patrol aircraft that recognises biological noise, from whales to shrimp, and removes it from operators' displays, backed by a secure shared environment into which the three nations contribute P-8 sensor data. Fielding has been uneven: as of June 2026 the UK Ministry of Defence told Parliament that the algorithms had not yet flown on RAF P-8s, although Britain draws on data and lessons from partner deployments.
In the United States the emphasis has moved from building a classifier to keeping it current. During RIMPAC 2026, Lockheed Martin and the US Navy demonstrated SensorMAX aboard MH-60R helicopters working sonobuoy fields; when a new contact appeared, operators labelled its signature, retrained the model in under five minutes and sent a small update back to the aircrew. The significance lies less in the algorithm than in the loop: the acoustic database stops being a static reference and becomes training data that is revised during an operation.
Britain's approach stresses rapid insertion. The Royal Navy's ASW Spearhead programme has given the Sonar 2087 towed array a new operator interface and uses a separate platform on which crews can trial new functions and revert to the original system, with AI and machine-learning detection explored in exercises alongside industry. The common thread across these programmes is spiral development driven by operator feedback rather than long, fixed procurement cycles.
This puts new weight on how acoustic information is managed. Labelled recordings, their provenance, the conditions under which they were collected and the versions of models trained on them all need to be stored, traced and, increasingly, shared between allies under strict release rules. In practice the information-management layer is becoming as important as the processing chain that sits on top of it.
For operators, the job is shifting from listening to supervising and curating. A sonar operator who labels a new signature is now also training the system, and an analyst reviewing a machine classification has to judge how far to trust it. Interfaces that show why a classifier reached its conclusion, display confidence honestly and keep the raw signal within reach will decide whether these tools reduce workload or simply move it elsewhere.

VIT Create UI

A visual effects studio turning its own craft into software that other people drive.
Real-time effects running off a camera — in a photo booth, or on a phone in someone's hand.
The company's advantage is the effects; the risk is that only the company can operate them. So the interface had to make an edit quick for someone who is not a compositor, and keep the software behaving the same way in places as different as a booth and a handset.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Real-time camera effects have moved from novelty to infrastructure in little more than a decade. Face filters on social apps taught a generation to expect a camera to alter what it sees, live, and the tools for building those effects grew into ecosystems of their own, with independent creators, brand campaigns and agencies that specialise in nothing else.
That ecosystem is consolidating, and the platforms are pulling in different directions. Meta shut down its Spark platform for third-party effects in January 2025, so that effects made by brands and independent creators disappeared from its apps while Meta's own remained. Snap moved the other way, widening who can author: in June 2025 it released Lens Studio apps for iOS and the web that let people build lenses from text prompts and simple editing tools, with the desktop application kept for professionals. For studios that build effects, the lesson is that a single platform is a fragile home.
Generative models are changing what an effect is. Classic augmented-reality effects track a face, hands or body and attach graphics to them frame by frame. Generative effects reinterpret the whole image, often on a server and after a short wait. Snap's sponsored AI lenses, launched in April 2025, generate personalised images of users with the company's own generative technology in place of hand-built 3D and visual-effects work. The two kinds of effect now sit side by side, and the line between a filter and an image-generation request is blurring, bringing questions of latency, processing location and consent that tracking-based effects rarely raised.
Physical venues show the same appetite in another form. In Korea the self-service photo studio has become a fixture of shopping streets. Life Four Cuts alone reports around 120 million visits since its 2017 launch and some 410 stores, with unstaffed booths that run around the clock, automatic retouching, output sized for social media, and limited-edition frames tied to K-pop idols and baseball teams. The booth has become a social ritual, and its effects are part of the reason people return.
For the people in front of the lens, expectations and anxieties are rising together. They expect an effect to respond at once, to flatter rather than distort, and to work in poor light and crowded frames. They increasingly need to know when their image is being reinterpreted by a model rather than decorated, where that happens and whether the result is kept. For those who build and operate effects, clear previews, easy ways to undo, and honest signals about what the software is doing to a face are becoming as important as visual polish.

OEM Navigation Map UI Design Specification Development

The specification is read by engineers building different cars for different countries. One document has to hold all of them.
A factory-fitted navigation screen is looked at while driving, and that fixes what may be on it. Beyond that nothing is fixed — model, region and year each move the specification.
So the real subject is not the map but how the variants are kept inside one document. How much is allowed onto the map, and where a difference is declared rather than duplicated.
Two-year contracts, revised against the development schedule. Working with the same organisation that long means the design principles are inherited rather than reopened, and what changes is written as change.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For most of the history of in-car navigation, the speed limit shown on the map was a convenience. In Europe it has become safety data. Under the EU's General Safety Regulation, every new vehicle sold since 7 July 2024 has had to carry intelligent speed assistance (ISA) and a drowsiness and attention warning, among other systems. A figure that once sat quietly in a corner of the navigation screen now drives warnings the driver cannot easily ignore.
Where that figure comes from turned out to be the difficult part. The map supplier TomTom explained in 2021 that ISA type approval requires speed-limit data from both cameras and digital maps. Cameras misread signs or have their view blocked, and many limits are implicit, applying on entry to a residential area, or apply only to certain vehicle classes and are never signposted. The map fills those gaps, and the camera catches what the map has not yet learnt. Map attributes that were once judged by the quality of a route are now judged by whether a warning is right.
Consumer testing is raising the bar again. The 2026 Euro NCAP protocols extend ISA checks from the test track to real roads, so the accuracy of speed-limit information is assessed in actual driving, and top ratings require driver monitoring with continuous eye and head tracking. The European Transport Safety Council, which welcomed the changes, also notes that voluntary ratings are no substitute for legislation and wants them reflected in the next revision of the regulation.
The consequences reach the screen. The limit drawn on the map, the one in the instrument cluster and the one the car acts on must agree, or the driver meets a contradiction at the very moment they are asked to trust the system. Conditional limits, by time of day, weather or vehicle type, have to be shown without crowding a display meant to be read in a glance. Signage conventions and units vary between countries, and so do the rules about which conditions apply. And because driver monitoring now records where the driver is looking, long glances at a busy map are measured rather than assumed.
For drivers, the change is subtle but real: the navigation display is no longer only advice. It is part of a system that intervenes, and people judge such systems quickly. A warning that is right most of the time but wrong on a familiar road teaches drivers to switch it off. Interfaces that show what the car believes the limit to be, make clear when that belief is uncertain and keep alerts proportionate will earn more trust than ones that simply sound louder.

b.IoT Smart Building Solution

Every release adds features. The person running the building is the same person, with the same amount of attention.
The energy solution shows a building manager consumption and cost. There is a great deal of data and only a few figures that change what anyone does.
So the priority is not what matters most but what has to be legible to someone who has just walked up. Which number leads, and how an abnormal reading makes itself seen without raising an alarm for everything.
We started from that — that nobody is watching — and re-ordered the information around the moment of return.
5 separate engagements on this line, 2018–2023.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
When Samsung unveiled b.IoT in Seoul in October 2017, the pitch was that sensors and camera feeds could switch heating, cooling and lighting according to whether rooms were actually in use. It was a familiar promise for the late 2010s, when building management was still largely the territory of a few controls specialists, each with its own wiring, protocols and head-end software. Over the years since, the category has changed character: from a supervisory console attached to the plant room towards a data platform that spans HVAC, lighting, metering and access.
Interoperability came first. BACnet and Modbus are decades old, but buyers increasingly treat open-protocol support as a condition of purchase rather than an option, because few estates run on one manufacturer's equipment. Samsung's own current product page leads with BACnet and Modbus interoperability, a 3D building digital twin and AI-driven savings for its VRF systems. That combination is now the industry norm: the large controls vendors all market cloud analytics and twin-style models layered over their controllers, and the differences between them lie more in data models and integration effort than in the headline features.
Regulation then turned an efficiency feature into an obligation, at least in Europe. The recast Energy Performance of Buildings Directive, adopted in 2024, requires new buildings to be zero-emission from 2030 and sets a path to a zero-emission stock by 2050. It also lowers the point at which non-residential buildings must carry automation and control systems: the Commission describes minimum requirements for systems above 290 kW by the end of 2024 and above 70 kW by 2029. Many mid-sized offices, schools and shops that once ran on time clocks and wall thermostats now fall inside that scope.
Artificial intelligence has arrived in a quieter form than the marketing suggests. Its most established uses are fault detection and diagnostics, forecasting of load and occupancy, and tuning start-up and set-back times against weather and tariffs. Digital twins have settled into a similarly practical role: less a photorealistic replica than a spatial index that ties meters, work orders and equipment histories to a floor, a zone or a single asset, so that the same building can be read by an energy analyst and a maintenance contractor.
For the people who run buildings, the result is more automation and a different kind of job. When software adjusts setpoints and schedules on its own, the operator's task shifts from issuing commands to supervising decisions made elsewhere. That puts weight on explanation: what the system changed, on what evidence, and how to override it without undoing the next optimisation cycle. Interfaces that make automated actions visible and reversible earn the trust that lets the automation do its work; those that hide them tend to be switched to manual and left there.

New Dreams Data Visualisation Consulting and Build

Executives do not read data. They look at it for as long as it takes to decide something, and then they stop.
A management report for a chief executive, on a tablet. Reporting had been split between paper and screens and multiplied into a different format for every department.
What a chief executive needs to see is not a smaller version of what a centre manager needs to see. So the work was to decide, level by level, which indicators appear at all — and how plain a chart has to be to survive a glance.
We interviewed eight people across three levels before drawing anything, then set the indicators for each level, brought the reporting formats into one, and laid out the daily, weekly, monthly and annual views on the principles that came out of it.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Business intelligence dashboards have been the standard way of putting numbers in front of senior managers for two decades, and the guidance on making them readable has changed little in that time. Nielsen Norman Group's advice, for example, is that charts should rely on length and two-dimensional position, the visual properties people judge quickly and accurately, rather than area, angle or colour intensity. Bar charts and plain line charts are read fast; decorative gauges and three-dimensional pies are not.
What changed from 2023 onwards is that the dashboard stopped being the only way into the data. Generative AI was folded into the major analytics platforms. Microsoft's Copilot for Power BI offers chat-based analysis, can summarise whole reports or particular topics, and helps analysts write calculation formulas. Salesforce's Tableau Pulse delivers personalised metrics and AI-generated explanations into the tools people already use, pitched explicitly at employees without a data background. The common idea is that a reader should be told what changed and why, in sentences, instead of having to find it in a grid of tiles.
That shift raises new questions. A written summary persuades in a way a chart does not, and it can be wrong while sounding certain. A language model summarising a dataset depends on the definitions underneath it being correct and agreed; if two departments count the same thing differently, the summary will repeat the inconsistency fluently. The unglamorous work of agreeing metric definitions and maintaining a shared semantic model becomes more important with AI, not less.
Consumption habits have shifted alongside. Rather than opening a dashboard on a fixed schedule, managers increasingly receive alerts when a metric moves outside its expected range, often on a phone or tablet. This favours a small number of well-chosen indicators with clear thresholds, and it raises the cost of noise: a feed that fires too often is ignored as surely as a dashboard nobody opens. Mobile layouts also force hard choices about what fits on a screen without scrolling.
For the people who read management information, the trend splits one task into two: noticing that something needs attention, and understanding why. Charts remain the quicker way to compare values and spot change. Generated text can help with explanation, but it needs to show its working, which figures it used, over what period and with what caveats, so that a decision-maker can rely on it or challenge it. Good reporting design now covers the sentence as well as the chart.

Donghae City Briefing Materials

The Donghae project was a plan by South Korea's Agency for Defense Development (ADD) to build a research institute for advanced unmanned systems on the coast at Donghae, Gangwon Province.
The Agency for Defense Development is South Korea's government agency for defence research and development, headquartered in Daejeon. Research on unmanned systems, particularly unmanned surface and underwater vehicles, typically relies on facilities with access to open water for trials, and the plan proposed such an institute on the east coast at Donghae.
AEON produced the presentation materials used by the Mayor of Donghae in briefings to residents on the plan.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Unmanned maritime systems are easy to draw and hard to prove. Autonomy software, sensors and data links have to be tested in real water, with real weather, currents and traffic, and that has made coastal test ranges and research clusters strategic assets in their own right. Countries that can run frequent, well-instrumented trials close to their engineers move faster from prototype to fielded capability.
Australia offers a clear example of a range used as an innovation hub. Jervis Bay on the New South Wales coast hosts the Royal Australian Navy's Autonomous Warrior exercises; the 2023 edition spent two weeks trialling autonomous undersea warfare systems on and below the surface. In 2024 the same waters were used by Australia, the UK and the US to test unmanned maritime systems under AUKUS Pillar 2, including the Ghost Shark extra-large autonomous undersea vehicle. The model of a controlled sea area, a naval host and open access for industry and allies shortens the loop between prototype and operational feedback.
South Korea has been building its own version of this infrastructure. In November 2025 ADD completed Donghaegwan, a research building in the Jinhae district of Changwon (despite the name, not in Donghae), dedicated to manned-unmanned maritime combat systems and unmanned surface and underwater vehicles, with laboratories for autonomy, data links, sensor fusion and combat-system integration. Weeks later DAPA, ADD and the Korea Institute of Ocean Science and Technology agreed to share test facilities and jointly develop test procedures and standards for maritime unmanned systems. The emphasis on shared standards matters: autonomous systems cannot be certified in the same way as conventional ships, and the methods for testing them are still being worked out.
Across these efforts the pattern is consistent. Governments co-locate research agencies, universities and companies near the water, invest in instrumented ranges and treat test data as a shared resource. Regions compete to host such facilities for the skilled jobs and industrial spin-offs they bring, while military users value the ability to run frequent small trials rather than rare set-piece tests. Civilian ocean science and defence research often share the same vessels, sensors and seabed infrastructure.
For the people who will operate these systems, test sites are where the human side takes shape. Early trials decide how much an operator is expected to supervise, how mission plans are handed to an autonomous vehicle, and what information comes back when communications drop. Ranges that put sailors alongside engineers from the start tend to produce systems whose interfaces reflect real watchkeeping rather than laboratory assumptions, and that difference shows when the systems reach the fleet.

Next Bixby UX for Advance UX Design

Bixby is summoned by voice. Its identity has to be a visible answer to something invisible.
It had to be recognisable on a watch, a fold, a vacuum cleaner, a television — and on machines with no screen at all.
The three states are the identity: listening, working, answering. They have to read the same whether a screen shows them, a light shows them, or a sound does.
We carried the existing identity forward into new visuals and motion, and wrote it as a guide that reaches devices with nothing to display on.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Voice assistants spent the late 2010s largely as command interpreters: they recognised a set of phrases and mapped each to a fixed action. Between 2024 and 2026 nearly every major assistant was rebuilt around large language models, and the change has altered not only what they can do but how they present themselves and where they live.
Amazon's Alexa+, announced in February 2025, is described by the company as a generative AI assistant able to hold open-ended conversations and act on requests such as controlling smart home devices or making reservations, on Echo speakers, Echo Show displays and Fire TV. A month later Google said Gemini would replace Google Assistant on Android phones and then move to tablets, cars, headphones, watches and home devices including speakers and televisions. The assistant is becoming a layer that follows a person across device categories rather than a feature of any one product.
Samsung is moving in the same direction. In February 2026 it introduced a new Bixby in One UI 8.5, recast as a conversational device agent that changes settings from plain descriptions, explains why a phone is behaving in a particular way and brings live web results into its own interface. Across the industry the emphasis has shifted from answering questions to operating the device on the user's behalf.
Visual identity has changed with it. Apple's 2024 redesign replaced Siri's familiar orb with a glowing light around the edge of the screen and added the option to type rather than speak. The assistant is drawn less as a character occupying one spot and more as a condition of the whole interface, which suits a system expected to act across apps and devices rather than return a single reply. Typing as an equal alternative to voice also acknowledges how often people are somewhere they cannot speak aloud.
The commercial logic is straightforward. Assistants that only answered questions were easy to ignore once search and chatbots did the same job better; assistants that can act inside the operating system have a role nothing else on the device fills.
For users, a more capable assistant raises the stakes of every exchange. When a spoken or typed request can change settings, send messages or place an order, people need to see what the system took from their words and what it actually changed, and to reverse it without effort. Confirmation before consequential actions, a visible record of what was done and a simple undo are what let people delegate to an assistant without having to watch over it.

Naval Engagement-Class Analysis Model (Build-I) UX/UI Development

The Naval Engagement-Class Analysis Model is a simulation model for analysing a single naval engagement; Build-I is its initial build.
An engagement-level analysis model simulates a single naval engagement in detail, representing the platforms, sensors, weapons and threats involved. In the usual hierarchy of combat models it sits between engineering models of individual equipment and mission- or campaign-level models of whole operations, and it is used to examine the effectiveness of weapon systems and tactics. Build-I is the initial build of this model.
AEON developed the UX/UI standards for the model's initial build.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Defence analysts usually describe combat models as a hierarchy. Engineering models represent a single sensor or weapon in physical detail; engagement models pit one or a few platforms against a threat; mission and campaign models aggregate whole forces over days or weeks. The engagement level is where physics meets tactics, and it is typically run many times over, varying geometry, timing and threat behaviour, to estimate how a weapon system or tactic performs across a spread of outcomes rather than in a single scripted case.
The tools behind this work are converging with those used for testing and training. BAE Systems' MIMESIS models seabed, weather and underwater acoustics together with ships, submarines and their combat management systems, sensors and effectors, and is offered for experimentation, for test and integration, where it provides digital twins of warfighting systems, and for training. In Britain, Dstl's Maritime Experimentation Lab at Portsdown West is a ministry-owned synthetic environment for trying out maritime command and control concepts, and in 2023 it sought development of a combat management system emulator with CMS-like operator displays and open interfaces such as DIS.
Artificial intelligence is the second shift. Research centres are experimenting with learning agents that can play out very large numbers of engagements, generating courses of action that human planners might not consider. At the US Naval War College, two students built a classified tool, recognised in June 2026, that uses statistical methods to estimate how likely different options are to succeed given each side's strengths and weaknesses. The common aim is reproducibility: results that can be rerun and compared rather than drawn from a single game.
The third shift is connection. Analytical models, training simulators and real combat systems increasingly share data standards, so that a scenario built for analysis can feed training and the reverse. US Fleet Forces and NSWC Corona recently showed the Chief of Naval Operations new capabilities added to the Navy's surface and air live, virtual and constructive training environments, linking sailors aboard ships with shore facilities. In this setting a digital twin is less a single product than a family of models of the same system at different levels of fidelity.
The weakness of all of this is credibility. A model is only as useful as its data and assumptions, and the more autonomous and interconnected a simulation becomes, the harder it is to see why a result came out as it did. Verification, validation and accreditation, and clear documentation of what each model does and does not represent, remain the slow, necessary work behind any headline about AI wargaming.
For the people who run these models, the interface is part of the method. Analysts have to set up scenarios consistently, read large result sets and explain them to decision-makers who did not build the model. Common conventions for how scenarios, entities and outcomes are presented make studies comparable across teams and over time, and tools that expose their assumptions rather than hide them are the ones commanders are most likely to trust.

CIWS(Close-In Weapons System)-II UI/UX Consulting

CIWS-II is a close-in weapon system under domestic development in South Korea for Republic of Korea Navy ships.
A close-in weapon system is the innermost layer of a warship's air defence, intended to destroy anti-ship missiles and other threats that get past longer-range defences. CIWS-II is being developed in Korea as the successor to the imported close-in weapon systems in Republic of Korea Navy service, combining a 30 mm Gatling gun with an active electronically scanned array radar, and is planned for future frigates and destroyers.
AEON provided UI/UX consulting for the CIWS-II system.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The close-in weapon system has always been the last layer of a warship's air defence, and recent combat has shown why it still matters. In January 2024 the US destroyer USS Gravely used its Phalanx to destroy a Houthi anti-ship cruise missile that had closed to roughly a mile, the moment in the Red Sea campaign when a US warship had to fall back on its gun. Most intercepts there happened far further out with missiles, but the episode was a reminder that outer layers can be penetrated and that the inner layer must be ready at all times.
The gun-based CIWS is itself changing. Phalanx and Thales' Goalkeeper, both Cold War designs built around radar-directed Gatling guns, are being replaced or supplemented. South Korea's CIWS-II pairs a 30 mm Gatling gun with an AESA radar drawing on KF-21 fighter technology and is intended as the last line of defence for the FFX Batch III frigates, KDDX destroyers and future large ships, with development scheduled at contract award to finish in 2027. The Netherlands chose a different answer for its Goalkeeper replacement, combining Rolling Airframe Missiles with guided DART rounds fired from 76 mm guns, both cued by Thales radar. Raytheon's SeaRAM, which puts RAM missiles on a Phalanx-style mount, sits between the two approaches. The gun-versus-missile debate for the inner layer has not been settled; navies increasingly carry both.
Drones have changed the arithmetic. Cheap one-way attack drones and uncrewed surface vessels, used in large numbers in the Red Sea and the Black Sea, make it unattractive to spend scarce interceptor missiles on every target. That has revived interest in guns with programmable ammunition and, above all, in directed energy. The UK has contracted MBDA to fit the DragonFire laser to a Type 45 destroyer by 2027, after trials in which it destroyed drones flying at up to 650 km/h, and the US Navy's HELIOS laser aboard USS Preble shot down a target drone in a test disclosed by the Pentagon's test and evaluation office. Lasers offer deep magazines at low cost per shot, but are limited by weather, power and dwell time.
The trend, then, is towards an inner layer built from several effectors, such as guns, short-range missiles, lasers and electronic attack, managed as one defensive system rather than as separate mounts. Sensors are moving to AESA radars and electro-optical trackers able to classify small, slow targets as well as sea-skimming missiles, and counter-drone duty is turning the CIWS from a weapon fired rarely in extremis into one that may engage repeatedly against swarms, including targets on the surface.
For operators, engagement timelines at this range are measured in seconds, so crews mostly supervise automatic modes rather than aim. As effectors multiply, the human task shifts towards choosing which weapon answers which target, keeping track of remaining missiles and laser availability, and holding the authority to permit or withhold fire. That only works if the system's classification is trustworthy and its intentions are visible at a glance, which makes the interface part of the weapon's effectiveness.

New Foldable Device UX Prototype Development

The first foldables were shipping, and every assumption an interface makes about its screen was suddenly negotiable.
A screen that changes size in the hand changes what was safe to assume — where things sit, what survives the fold, what should happen at the moment it opens.
Much of that can only be judged with the thing in your hand, so the decision was which scenarios were worth building rather than describing. Interface, graphics and visual identity were resolved together into a prototype that actually ran.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
When Samsung's first Galaxy Fold reached reviewers in April 2019, the story was durability: some early units suffered display failures and the launch was postponed, with the device finally going on sale that September. Seven years later the conversation has moved on. Foldables are a settled product category with several generations behind them, and the work of making software fit them has shifted from individual apps to the platforms themselves.
Google's first large step was Android 12L in 2022, a release aimed specifically at tablets and foldables that reworked system screens for large displays, added a taskbar with drag-and-drop split screen and improved how unoptimised apps appear. The more lasting contribution was conceptual. Android's guidance now steers developers towards window size classes, breakpoints based on the space an app actually has rather than the device it runs on, and says plainly that they are not meant for checking whether a device is a tablet. On a foldable the available window can change while the app is running, through folding, rotation or multitasking, so a layout tied to device type will be wrong at some point in the session.
The web has followed. The W3C's Device Posture API, a Candidate Recommendation Draft implemented in Chromium-based browsers, lets a page know whether a device is lying flat or folded and respond through script or a CSS media feature, for instance by placing content on either side of a hinge.
The hardware has meanwhile stretched towards tablet size. Huawei's Mate XT introduced the tri-folding phone in 2024, and Samsung's Galaxy Z TriFold uses two hinges to open to a 10-inch main screen. This month Apple entered the category: its iPhone Duo, announced on 15 October, pairs a large inner display with a 5.4-inch outer one, and Apple's description stresses content that responds as the device folds, Split View on the inner screen and StandBy on either display.
Apple's arrival matters less for the hardware than for the software ecosystem. Developers who treated foldables as a minority Android concern now face the same problem on both major mobile platforms: interfaces that must stay coherent while the screen beneath them changes shape, size and orientation, sometimes in the middle of a task.
For users, the promise of a foldable is continuity, starting something on a narrow outer screen and carrying on with more room when the device opens. That only works when the transition keeps a person's place and state, when controls stay within reach of a hand holding an unfamiliar shape, and when apps make real use of the larger canvas rather than simply stretching. A good foldable interface is one in which the user rarely has to think about the fold at all.

Next-Generation Dreams UX Consulting Development

The teachers had done this on paper for decades. The system was newer than the method and worse at it.
Home-visit tutoring still ran on paper. A teacher on their round, a branch director, a regional manager, head-office staff and executives — five different days, one system, and it served none of them well.
A tool that loses to paper is not losing on features. So the question was what each role actually does in the course of a day, and how plain the screens have to be for people who have never needed them to be anything else.
We ran focus groups on the work itself and on their experience of the system, and proposed a UX direction and baseline guidelines out of what came back.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Software for people who work away from a desk, such as field engineers, care workers, sales representatives and tutors on home visits, has gone through several phases since the first rugged handhelds. Tablets and smartphones made the hardware cheap, and cloud services made the data shared. One observation has nevertheless proved durable: paper is far harder to displace than technology plans assume. In The Myth of the Paperless Office, published in the early 2000s, Abigail Sellen and Richard Harper explained paper's persistence through its affordances, the physical qualities that make it easy to carry, fold, mark up and share, and argued for paper and digital tools working together rather than for the elimination of one.
Field software has since developed specific answers to those strengths. Active styluses brought handwriting back to tablets. Offline-first design became a standard pattern on the web as well as in native apps: Google's web.dev guidance sets out techniques for building applications that keep working without a network, using service workers and local caches, which matters to anyone working in basements, lifts or rural areas. Low-code platforms let organisations adjust forms and workflows without waiting for a major release.
Method has changed as much as technology. Public-sector digital services set a reference point: the UK government's Service Standard begins by requiring teams to understand users and their needs before settling on a solution. In enterprise software that has meant more research where the work actually happens. Nielsen Norman Group describes field studies as research carried out in the user's own setting, where observation exposes workarounds and constraints that a meeting-room interview tends to miss. For mobile roles the unofficial workflow, a notebook in the car, a message thread among colleagues, a spreadsheet kept at the local office, is often the real system.
The newest wave is AI assistance for frontline staff: summarising a customer's history before a visit, drafting notes from speech, or proposing the next task. These features suit people who spend the day moving between locations, but they inherit an old problem. If the underlying record is thin because entering data is a chore, the assistant can only summarise the gaps. Adoption still depends on whether the basic act of recording is quicker than the method it replaces.
For workers in the field, a tool is judged in seconds and in context: one hand busy, a customer waiting, a weak signal. The lesson of the past decade is that such users rarely resist technology as such; they resist overhead. Software earns its place when it removes steps people already find tedious, respects routines that already work, and gives something back, such as information they could not get from paper, in return for what it asks them to record.

Home Appliance Device Mobile App UX/UI Design

Buy a different appliance and you should not have to learn the app again.
When every product is controlled its own way, the customer pays for the inconsistency. The work was finding the frame that holds across product lines — the flow and the interface for the appliance app, designed on a structure the categories share.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For most of the past decade, buying a connected appliance meant installing its maker's app. A household with a refrigerator, a washer and an air conditioner from three brands ended up with three apps, three accounts and three ways of saying that a cycle had finished. Since about 2022 the industry has been trying to undo that on two tracks at once: a device-level standard that any platform can speak, and agreements between appliance makers to let their clouds talk to each other.
The device-level track is Matter, the standard run by the Connectivity Standards Alliance, which launched in late 2022 with lights, locks and thermostats. Samsung added Matter control to its SmartThings app that December and supports multi-admin, so the same device can be controlled from several platforms. Appliances followed a year later. Matter 1.2 added refrigerators, room air conditioners, dishwashers, laundry washers, robot vacuums and air purifiers, each defined by a deliberately limited set of functions: remote start, progress notifications, a handful of modes and error alerts. The standard does not try to describe everything a machine can do. It describes what every app should be able to do with it.
That common subset is already visible in platform apps. When Apple's Home app gained support for Matter robot vacuums in iOS 18.4, it offered start, stop, cleaning presets and return-to-dock, while room mapping still required the manufacturer’s own app. The pattern is becoming general. Platform apps handle the fundamentals across brands, and the maker's app keeps the depth: detailed programmes, maintenance and diagnostics.
The cloud track matters most for large appliances, which mostly connect over Wi-Fi to their makers' own services. Through the Home Connectivity Alliance, an industry group founded in 2022, Samsung and LG began connecting SmartThings and ThinQ in 2023 so that either app could operate the other’s appliances, with energy management planned for a later version of the specification. For two rivals to expose their machines in each other's apps would have been hard to imagine in the mid-2010s. It shows how far the expectation of a single place to manage the home has travelled.
The result is a layered arrangement rather than a single winner, and it changes what people experience. The same washer may now appear in a platform app, in a rival's app and in its own, each drawing it slightly differently. For users, the important things are that a cycle, a state or an error is called the same thing wherever it appears, that basic actions behave identically across apps, and that the reason to open the maker's app is the depth it offers rather than the lack of anywhere else to go. Designers increasingly have to assume that their device will be shown through someone else's interface, and to make its names and states clear enough to survive the translation.

SmartThings iOS GUI

The same service on two platforms. Some of the difference is required and some of it is only drift.
People move between an Android phone and an iPhone and expect to find the same service on both.
So the work was separating what must match from what must differ. One list is about identity and the other about convention, and confusing them is how an app ends up feeling wrong on one platform and generic on the other.
We set the Android reference design and the iOS guidelines side by side, wrote down the differences, and built the graphics from that.
2 separate engagements on this line in 2022.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For much of the 2010s, the two mobile platforms' design languages seemed to be drifting toward each other: flat surfaces, similar navigation patterns, comparable type scales. Since around 2021 the opposite has happened. Google and Apple have each reasserted a distinctive visual identity, and the distance between an Android app and an iPhone app that follow their platforms' guidance faithfully has grown rather than shrunk.
Google's turn came with Material You, introduced with Android 12 in 2021, which extracts a colour palette from the user's wallpaper and applies it across the interface. It made the owner's device, rather than the app's brand, a source of colour, and pushed app makers to decide how much of their own palette to keep. In 2025 Google followed with Material 3 Expressive, which it says draws on 46 research studies with more than 18,000 participants; its eye-tracking work reported key controls being spotted markedly faster in the more expressive designs, while conceding that departing from familiar patterns can hurt usability.
Apple answered in June 2025 with Liquid Glass, a translucent material that refracts the content behind it and adapts to context, which Apple described as the first design to extend across iOS, iPadOS, macOS, watchOS and tvOS at once. Reception among usability specialists has been mixed. Nielsen Norman Group criticised its legibility over busy backgrounds, its controls that appear and collapse unpredictably, and the relocation of long-standing elements such as search.
Underneath, tooling has moved the other way. Cross-platform frameworks such as Flutter, React Native and Compose Multiplatform have matured to the point where a large share of an app's interface code can be written once for both platforms. That has lowered the cost of building for two ecosystems, while turning platform fidelity into a deliberate choice rather than something that comes for free with native development.
For users, many of whom switch between ecosystems over the years or carry one of each, these changes mean their expectations are shaped by two increasingly different visual grammars: where search lives, how controls respond to touch, what colour is allowed to signal. Interfaces serve them best when the behaviours people rely on without thinking, such as navigation, gestures, system dialogues and accessibility settings like reduced transparency and larger text, follow the platform in their hand, and when visual expression never comes at the cost of contrast or predictability.

AI Object Recognition Training Simulator UX Design

An AI object recognition training simulator is a system for training neural-network models to detect and classify objects, in this case by reinforcement learning running on a neural processing unit (NPU).
Object recognition models used in defence must learn to identify vehicles, vessels, aircraft and people from sensor imagery under widely varying conditions. Simulators allow such models to be trained and evaluated against generated scenes, and reinforcement learning lets a model improve through repeated interaction with a simulated environment. NPUs are processors designed to run neural-network workloads efficiently, which suits them to embedded and edge applications.
AEON carried out the UX design for a simulator for training object recognition by reinforcement learning on an NPU.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The war in Ukraine has turned onboard machine vision from a research topic into a production requirement. Electronic warfare routinely cuts the link between a drone and its operator, so the drone itself must be able to recognise and hold a target. As of September 2026, Ukrainian forces were reported to be using more than 70 systems that apply AI and computer vision to navigation, object recognition and target engagement, although a presidential adviser warned in the same report that Russia had pulled ahead in running neural networks directly on board its drones.
Western products reflect the same priorities. Helsing's HX-2 strike drone uses onboard AI for camera-based navigation without satellite positioning, for automatic target recognition, and for terminal tracking that continues if the data link is disrupted, while the company says the engagement decision stays with a human. Autonomy for perception and guidance, with human authority over the use of force, is becoming the default Western design stance.
Running these models on a small airframe or a soldier-worn device is a hardware problem. General-purpose GPU modules have carried most of the load so far, but dedicated neural processing units promise comparable inference at lower power and heat, which matters wherever batteries and cooling are scarce. Korea has an active NPU sector, and in 2026 DEEPX said its chips would support AI-powered perimeter surveillance at Republic of Korea Air Force bases, an early sign of domestic edge-AI silicon moving into defence use.
The other bottleneck is data. Real imagery of military targets in every condition is scarce, sensitive and costly to label, so forces increasingly train on synthetic scenes. The US Air Force, for example, used an AFWERX contract to have Applied Intuition generate synthetic data to train an aided target recognition algorithm. Simulation also enables reinforcement learning, in which a model improves by acting in a virtual environment rather than by studying labelled examples. The open question is transfer: models that excel in simulation can falter against real sensor noise, camouflage and decoys, which keeps live testing essential.
For operators, on-device recognition changes the job from finding targets to judging the machine's findings. The interface has to convey what the model believes, how confident it is and, where possible, why, at a glance and under stress. Training matters as much as the interface: crews need to see where models fail, not only where they succeed, so that trust is calibrated rather than assumed. Simulators that expose both the AI and its human supervisor to the same difficult cases are likely to become as important as the models themselves.

NHN Cloud Training Centre UX Planning

A person arrives knowing what they want to learn, not what the catalogue calls it.
The centre ran a great many courses at a great many levels. Classification is the whole design here — how the courses are divided decides whether anyone finds theirs.

DICEVIEW UX Development

The endoscope is swallowed. From then on the doctor is working something they cannot see, in a room built around a machine the size of a scanner.
A capsule endoscope is not a smaller endoscope. The doctor drives it with a control in one hand while reading live video with the other, and the whole procedure has to be recorded precisely as it happens.
The design question is the room, not the screen — when the view has to change, and how the patient's data moves between the instrument and the record without anyone entering it twice.
We looked at the space and the procedure before drawing anything, then designed the graphics for the monitoring software.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Capsule endoscopy has been in clinical use since the first device was cleared in the United States in 2001, and for most of that time its strengths and weaknesses stayed the same. A swallowed camera can reach parts of the small bowel that a conventional scope cannot, without sedation. But the capsule moves passively with the gut, and it returns hours of footage that a specialist must review frame by frame. The past several years have brought sustained work on both of those limits.
Reading time was the first target for machine learning. A widely cited study in Gastroenterology in 2019 trained a convolutional neural network on small-bowel capsule images and reported that AI-assisted reading reached about 99.9 per cent sensitivity against roughly 75 per cent for conventional reading, while cutting average review time from about 97 minutes to about 6. Results of that kind have since been tested in many settings, with more modest figures in some, and AI tools for the digestive tract have moved into regulated products. The US Food and Drug Administration's list of AI-enabled medical devices now carries a growing set of gastroenterology entries, among them software named Deep Capsule cleared in March 2026.
Movement was the second target. In May 2020 the FDA granted a De Novo authorisation for the NaviCam system and created a new classification for magnetically steered capsule endoscopy. An external magnet guides the capsule through the stomach, turning a passive recorder into an instrument that can be directed at a region of interest. Steering brings the examination closer to conventional endoscopy in what it can see, while keeping the patient free of a tube and, in most cases, of sedation.
Taken together, these developments change where the work happens. Reading can be partly automated and done remotely; the examination itself can be planned and adjusted in real time rather than simply recorded. Research groups and manufacturers are also exploring capsules for the colon and for wider screening programmes, where the capacity of endoscopy units is a persistent constraint. The open questions are less about whether the technology works than about how it fits into referral pathways, staffing and quality assurance.
For clinicians, the shift is from watching to verifying. When software proposes which frames matter, the reader's task becomes judging candidates and remaining alert to what the model might have passed over, a setting in which automation bias is a known risk. Interfaces that show why a frame was flagged, make it easy to review surrounding context and track what has and has not been inspected help keep human judgement engaged. For steered systems, the familiar challenges of remote manipulation, including orientation, feedback and fatigue, come to the fore.

SmartThings 3.0 UI Guidelines

Several smart-home apps had just been folded into one. Everything arrived at once and none of it had been introduced.
Products and functions that had lived in separate apps now shared a single structure, and the structure showed the strain.
The guideline had to decide what all these devices have in common before it could decide how they look. A flow and an information structure that hold whether the thing on the other end is a television or a door lock.
The 3.0 guidelines set the rules across the whole app. Having worked on it since the pilot, we knew why the earlier decisions were made and could move them rather than guess at them.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In 2020 the smart home was still divided along ecosystem lines. A device might work with Amazon Alexa, Google Assistant, Apple HomeKit or Samsung SmartThings, but rarely with all of them, and each platform kept its own model of what a light, a lock or a thermostat could do. Buying a device meant first checking a grid of logos on the box.
The industry's response took shape over the following years. In 2021 the Zigbee Alliance renamed itself the Connectivity Standards Alliance, and in October 2022 it released Matter 1.0 along with a certification programme and an open-source reference SDK. The first release covered lighting and electrical products, HVAC controls, blinds and shades, safety sensors, door locks, TVs and other media devices, controllers and bridges, running over Wi-Fi, Ethernet and Thread, with Bluetooth Low Energy used to bring devices onto the network. The definition of what a device type is and what it can do began to live in a shared specification rather than inside each platform.
Later releases widened that shared vocabulary. Matter 1.4, published in November 2024, added device types for solar power, batteries, heat pumps and water heaters, along with EV-charging preferences and energy-aware scheduling. It also introduced an enhanced multi-admin feature, so that a single act of consent can connect a device to several ecosystems at once, and a certification category for home routers that double as Thread border routers. The smart home's centre of gravity was shifting from convenience toward energy management.
Platforms have adapted around the standard rather than disappearing. Samsung now says most Matter devices work with SmartThings out of the box, through hubs built into recent TVs, monitors, refrigerators and soundbars, and through other brands' hubs acting as Matter bridges. Manufacturers have also formed the Home Connectivity Alliance, established in 2021 with the stated aim of breaking down walled gardens between brands. Adoption has been gradual, but the direction is no longer in doubt.
For residents, standardisation means the logic of the home increasingly lives in shared definitions rather than in a single vendor's app. Setting a device up becomes a matter of scanning a code, and the same lock can appear in several apps at once. That raises interaction questions of its own: which app is the authority when two ecosystems disagree, how a household learns who else has access to a device, and how controllers present the same type of device consistently whatever brand sits behind it.

Hyundai MN Navigation UI Design

Features keep arriving. The menu they arrive into was settled years ago.
A premium navigation system carries a great deal and appears in many different cars. The driver glances and decides.
So the real work is finding where a new function fits without moving anything the driver already knows. An extension that disturbs the screen's rules costs more than the function is worth.
A yearly engagement, revising the specification alongside the planning team. Each time a function was added we checked first where it collided with the existing rules.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In 2020 a car's navigation system was, in most cases, still a product of its own: the carmaker's software, updated rarely, sitting beside a phone that did much of the same work. Since then the two have converged from both sides. On the carmaker side, Android Automotive OS has become a common base. Unlike Android Auto, which projects a phone's apps onto the car's display, it runs directly on the vehicle's own infotainment hardware, and manufacturers can choose to license Google's apps and services on top of it.
On the phone side, Apple has pushed deeper into the car. CarPlay Ultra, which began rolling out with Aston Martin in May 2025, supplies content for every screen in front of the driver, including the instrument cluster, and reaches vehicle functions such as climate and radio as well as car-specific settings. The phone platform is no longer a guest window inside the carmaker's interface; it can be the interface.
Hyundai's own answer arrived in 2026. Its Pleos Connect system, which debuted in the Grandeur in Korea in May, is based on Android Automotive and adds Gleo AI, an assistant built on a large language model that handles navigation, climate, web searches and the owner's manual and can process several commands in one request. A third-party app market lets drivers install software, with services such as YouTube and Spotify running on the car's screen without a phone. Buttons on the wheel and below the display remain for common functions.
Together these changes alter how new functions reach the driver. In a closed head unit, the manufacturer alone decided where everything lived. On a platform, functions arrive as apps, as voice requests or as panels competing for space on shared screens, often built by developers who have never designed for a moving vehicle. Navigation becomes one service among several, sometimes from the carmaker, sometimes from a phone ecosystem, sometimes both in the same car. What holds the experience together is increasingly the platform's layout rules, its driving-mode restrictions and its assistant.
For drivers, the gain is familiarity and choice: the apps and accounts they already use follow them into the car, and a spoken request can replace several taps. The cost is consistency. When several parties shape what appears on the screen, the same task can work differently from one app to the next. The design effort shifts towards rules that every contributor must follow, and towards an assistant that can say, in plain words, what it has just changed in the car.

Continence Care Informatisation Service UI/UX and Development

A care device worn on the body, and a record of it read by a doctor, a care worker and a family member. It cannot mean the same thing to all three.
The robot has its own controls on the patient's side; everything else happens on a tablet or a desktop, at the bedside, at the station and at home — and a good deal of it in the middle of the night. The patient's condition and their rhythm have to be watched continuously, which is a great deal of data arriving from a private place.
So the first structure to settle was not the screen but who is allowed to see what. Where the clinical view divides from the lay view, and whether the care log belongs in the same flow as the alerts and the statistics.
We separated the information by stakeholder before designing, then built the permissions from that. One adaptive layout covers desktop, tablet and phone, and a dark mode was designed alongside it because so much of this happens at night.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Population ageing is reshaping continence care faster than most areas of nursing. The World Health Organization expects that by 2030 one person in six worldwide will be aged 60 or over, and lists urinary incontinence among the geriatric syndromes common in later life. South Korea crossed the threshold of a super-aged society at the end of 2024, with more than a fifth of residents aged 65 or over, and much of East Asia and Europe is following a similar path. Continence is one of the most frequent reasons older people need help, and one of the most sensitive.
Clinically, the emphasis has moved from containment towards management. WHO's guidelines on integrated care for older people name urinary incontinence among the age-associated conditions that community services should actively manage, rather than treat as an unavoidable part of frailty. Its Decade of Healthy Ageing, running to 2030, counts access to good-quality long-term care that lets people live with dignity among its four action areas.
Technology has entered this space from several directions. Sensor-equipped pads record wetness and timing so that changes can follow an individual's pattern rather than fixed rounds. Wearable ultrasound devices estimate bladder fullness and prompt a timely trip to the toilet. Automated excretion-care devices handle cleaning and collection for people confined to bed, and Japan has included excretion support among the priority areas of its national programme for robotic care equipment. Many of these products began in care homes and are now being adapted for use at home, where family members take on part of the work.
These tools belong to a broader category that is growing quickly. WHO estimates that more than 2.5 billion people already need at least one assistive product, a figure expected to rise with population ageing. At the same time, many countries report shortages of care workers, which strengthens the case for technology that saves unnecessary checks and lets staff spend their time where it is needed. Evidence on outcomes is still building, and the most persuasive results tend to come from programmes that combine devices with changes in care routines rather than from devices alone.
For the people involved, the design questions are as much about dignity as about efficiency. An older person should be able to understand and, where possible, decide what is being monitored. Caregivers, paid or not, need prompts that respect their workload rather than adding to it, and information that supports a conversation with the person rather than replacing it. Interfaces in this field succeed when they make care quieter, more timely and more personal, and fail when they turn an intimate need into a stream of notifications.

L-SAM Operations/Engagement Control Centre UX

L-SAM is South Korea's long-range surface-to-air missile system, forming the upper layer of the country's indigenous air and missile defence.
L-SAM (Long-range Surface-to-Air Missile) is a South Korean air and missile defence system developed by the Agency for Defense Development together with domestic industry. It is designed to intercept ballistic missiles at altitudes above 40 km as well as aircraft, and forms the upper layer of the Korea Air and Missile Defense (KAMD) architecture, above the Patriot and Cheongung-II systems. Development was completed in 2024.
AEON provided UI/UX design for the multifunction console in the system's operations and engagement control centre.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Upper-tier interceptors get the headlines, but the place where an engagement is actually run is the battle management centre, and it deserves as much attention as the missile. The Arrow system set an early template. Its Citron Tree centre, as described by Army Technology, fuses Green Pine radar tracks with other data, runs intercepts automatically against single or multiple threats, lets operators step in at every stage, and plots the predicted impact area and best intercept point on an electronic map. Automation by default, with a human able to override, remains the basic contract in most upper-tier systems.
What is changing is how far these centres reach beyond their own battery. In the United States, the Army has been working to bring THAAD under its Integrated Battle Command System rather than leaving it as a separate fire control island. In early 2024 it planned a test to push data through IBCS into THAAD's command and control and measure how much the interceptor's systems could absorb. The question is not only technical. Once an upper-tier battery can take tracks from any sensor on the network, its operators are no longer looking at one radar's picture but at a fused one, with all the questions of track quality and ownership that brings.
Europe is following a similar path with its own hardware. The SAMP/T NG programme gave the battery a new engagement module, the command element that links radar and launchers and can act either as a coordination node or as the heart of a fire unit. In December 2025 the system fired as a single fire unit in both its Italian and French configurations, at Salto di Quirra in Sardinia and at Biscarrosse, with Link 16 among its supported data links.
Across these programmes a few common directions are visible. Consoles are becoming general-purpose, multifunction workstations whose role is set by software, so that one position can serve surveillance, engagement or coordination as the situation demands. Software architectures are opening up, so that new sensors and interceptors can be added without rebuilding the operator environment. Track counts are rising with modern AESA radars, which shifts the burden from finding targets to deciding which of many deserve attention. And engagement timelines against ballistic targets remain short enough that the machine proposes and the crew disposes, often within seconds.
For operators, this means the console is less a window onto one weapon and more a seat in a network. The skills that matter shift towards judging automated recommendations, understanding where a fused track came from and how far to trust it, and keeping a coherent picture when several systems report the same object. Good human-machine interaction in this setting is about making the system's reasoning legible quickly, so that intervention is possible when it is needed and unnecessary when it is not.

MDE Base Service UX Development

Multi Device Experience — the promise that what you were doing on one device is already waiting on the next.
A handover is only believable when it is invisible. The design question is less the moment of transition than what the second device already knows by the time you pick it up — and how much of that has to be settled before the first one is put down.
This was the base service in the sense the word carries here: the first one, written so the rest could be cut from it. A phone, a tablet, a television and a computer do not divide the work the same way, so what gets decided at the centre is not a screen but the part all of them have to agree on.
Writing something to be inherited is a different job from designing it. A choice that is merely convenient at the centre becomes expensive at every edge, and it reaches each device team as a constraint rather than a suggestion.
UX development for the service across the range.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Moving work between devices has been a promise of every large ecosystem for a decade, but its form keeps changing. Early continuity features passed a single task from one screen to the next. The current generation increasingly lets one device take direct control of another, and pressure from regulators and industry partners has begun to open these features beyond a single brand.
Apple's macOS Sequoia, released in 2024, added iPhone Mirroring, which shows and operates a locked iPhone in a window on the Mac, with keyboard and trackpad control, drag-and-drop file transfer and iPhone notifications answered from the desktop. Samsung's Multi Control takes a related approach between Galaxy phones, tablets and Galaxy Book laptops, letting one keyboard and pointer move across them. The emphasis has moved from handing a task over towards treating several devices as one workspace.
Sharing has been consolidated too. At CES 2024 Google and Samsung merged Nearby Share and Quick Share into a single Quick Share service across Android phones and Chromebooks, with Windows PCs added through manufacturers, and with discovery settings that let people choose between everyone, their contacts or only their own devices. Two overlapping tools on the same platform had been a steady source of confusion; one name and one behaviour removed it.
Regulation is now pushing the boundary further. In March 2025 the European Commission specified interoperability measures under the Digital Markets Act requiring Apple to open nine iOS connectivity features to other manufacturers, among them showing iPhone notifications on third-party smartwatches, peer-to-peer Wi-Fi transfers and easier pairing. Continuity that was once a reason to stay inside one ecosystem is, at least in Europe, becoming something devices from other brands can expect to share.
The number of personal devices keeps growing alongside these changes. A typical household may now combine phones, tablets, laptops, watches, earbuds and televisions from more than one company, and cross-device features are judged less by what they can do in a demonstration than by whether they work every time on the hardware people actually own.
For users, the value of continuity depends almost entirely on predictability. People need to know which device is in control, where a file or call will land, and who nearby can see their device. When those questions have obvious answers, several screens feel like one tool; when they do not, they feel like several tools competing for attention.

Samsung PC Software UX

Software nobody chose. It is already on the machine when the box is opened, and it has about a minute to earn the space it takes.
Samsung builds the PC and ships it with its own software on it. Samsung Flow and SideSync for moving between a phone and the machine; PC Settings, Samsung Update and Smart Manager for the machine itself; the gallery, the recorder, the camera control, the radio; Samsung Security; Air Command for the pen; a companion for the television; a layer for the Odyssey gaming models. Around two dozen titles. To the person opening the box they are not two dozen titles — they are what the computer is like.
Preinstalled software is judged before it is used. None of it was searched for or downloaded, so each one has to explain itself on its first screen or be removed, and removing it is easy. That is the brief: the shortest distance between opening the thing and it being obviously worth having open.
The other half is that they have to read as one manufacturer rather than a dozen teams. A settings panel and a photo gallery have almost nothing in common as tasks, but the same person meets both on the same machine within the same hour. What holds them together is not a style but a set of decisions repeated — where things live, what a state looks like, how far the software may go in interrupting.
UX, UI and GUI across the suite, sustained through versions — SideSync to SideSync 4.0, Samsung Update to 3.0, PC Studio Plus to 1.5. 10 separate engagements on this line, 2015–2020.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, the software PC makers install before a machine leaves the factory has been squeezed from three directions: regulators, the operating-system vendor and a change in where manufacturers believe their value lies. In 2015 the preinstalled layer was often a mix of trial offers, update agents and utilities, and its reputation was poor. A decade on, it is narrower in some respects and considerably more ambitious in others.
Trust was the first pressure. In 2017 the US Federal Trade Commission settled with Lenovo over advertising software preloaded on laptops from 2014 that intercepted encrypted web traffic to inject pop-up ads. The order required affirmative consent before such software is preinstalled and a security programme, subject to outside audits, for most of the consumer software Lenovo preloads, lasting twenty years. The case became a reference point across the industry: anything shipped on a machine by default carries the manufacturer's name, whoever wrote it.
Regulation then reached the operating system itself. To comply with the EU's Digital Markets Act, Microsoft made more built-in Windows apps removable, including, in the European Economic Area, the Edge browser and Bing web search. Once even the platform owner's own applications became optional, the expectation that any preinstalled component can be uninstalled hardened from courtesy into norm.
Manufacturers' software ambitions, meanwhile, moved into the spaces between devices. Samsung's recent Galaxy Book laptops emphasise Multi Control, Quick Share, Second Screen and Storage Share, which let a phone, tablet and PC share input, files and screen space. Apple's Continuity features and comparable efforts from other vendors point the same way: the distinctive part of a manufacturer's software is increasingly how it connects the devices a person already owns.
The latest pressure comes from on-device AI. In 2024 Microsoft introduced Copilot+ PCs, a class of Windows machines with neural processors that run features such as live caption translation locally, with Samsung among the launch manufacturers. The same launch showed the risk of ambitious defaults: Recall, which saved snapshots of on-screen activity, was held back for testing with Windows Insiders, with Windows Hello sign-in required to use it, after security researchers raised concerns.
For the people who buy these machines, factory software now sits within a clearer set of expectations than it did a decade ago. Designing it well depends on consent and control as much as on features: asking before collecting anything sensitive, being honest about what each feature does with personal data, making removal straightforward, and letting cross-device features stay out of the way until someone reaches for them.

SmartThings Home Appliance GUI,UI Design and Guide

Appliances were joining faster than screens could be drawn for them.
Each product brought its own control screen, and the number was climbing.
Consistency at that speed is a matter of writing the guide while drawing, not after. Otherwise every product settles its own answer and the guide arrives only to describe the mess.
Product screens and the guide were made together, and the scope grew during the work.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Around 2020, the main job of a connected-home platform was to give each device a usable control screen in the app and to let people switch things on and off from a distance. In the years since, the emphasis has shifted from controlling devices one at a time to coordinating them. Hubs have moved inside ordinary appliances, automation has become something people describe rather than configure, and energy has become a concern the whole household shares.
The hub was once a separate box that enthusiasts bought and paired. At CES 2022 Samsung announced that SmartThings hub software would be built into selected TVs, monitors and Family Hub refrigerators, with support for Matter planned, removing the need for a dedicated device. The refrigerator that once needed its own control screen in an app could now host the connections for the lights, plugs and sensors around it. The question of where a device's interface lives became less tidy: on the phone, on the device, or on whichever large screen happens to be nearby.
Automation has gone through a larger change. Writing routines used to mean assembling triggers, conditions and actions in nested menus. Google's Gemini for Home, launched in October 2025, lets people create automations by describing what they want in a sentence, which Google itself framed as taking the feature out of menus built for enthusiasts. The interface task moves from building a rule to checking an interpretation: did the system understand what was meant, and will it do what the person expects at three in the morning?
Energy has given coordination a practical purpose. SmartThings Energy, which Samsung describes as the first platform of its kind to earn ENERGY STAR certification as a smart home energy management system, added automatic participation in utility demand-response programmes and an AI energy mode for appliances in 2023. Standards are moving the same way: Matter 1.5, released in November 2025, defines how devices can receive tariff and grid carbon-intensity data in a common format and adjust their operation. A washing machine is no longer only something to start and stop. It becomes a participant in decisions made for the whole house.
For the people living with these systems, the useful question is less about finding the right control screen and more about understanding what the home has done and why. A wash moved to an off-peak hour, a routine that fired unexpectedly, or a device switched off by an energy event all need a plain explanation, a preview before they take effect where possible, and an override that is easy to find. As the system takes on more of the doing, the interface takes on more of the explaining.

Hanwha Precision Machinery - Welding Robot SW Resource Artwork Development

A collaborative robot is sold on the idea that the person on the floor can set it up themselves. The programming screen is where that claim gets tested.
Hanwha's industrial collaborative robots are driven from a single panel that controls the cell. Before the robot does anything it has to be taught the job, and in the software as it stood that teaching had been cut into steps that each demanded a decision. The operator was making a great many selections and entering a great many values before the first weld happened.
That cost is easy to overlook because it is paid before the work starts, and it falls on someone whose trade is welding rather than robotics. Each input is also a place to be wrong, on a machine that then moves.
So the work was to take steps out rather than to make them look better. Which decisions the software could reach on its own, which ones belong together on one screen, and which had only ever been separate because that is how the code had grown. Interface, screen design and graphics for the control panel.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Welding has become one of the clearest cases for collaborative robots, and the reason is people as much as machines. Experienced welders are retiring faster than new ones are trained; in the United States the AWS Foundation, affiliated with the American Welding Society, projects that more than 300,000 new welding professionals will be needed by 2029. For fabricators that cannot hire enough skilled hands, a robot that shares the floor with the people they do have is an attractive proposition.
Cobots grew up as the option for smaller shops: lighter arms, less fencing, and teaching by hand-guiding instead of specialist programming. Korean industry took part early, and Hanwha spun its collaborative robot and factory automation businesses out into a separate company, Hanwha Robotics, in October 2023. Growth has since become steadier rather than explosive. The International Federation of Robotics reports that cobots accounted for about 11% of industrial robot installations in 2025 and grew more slowly than the robot market as a whole, a sign that the easy early applications have been taken and the remaining ones demand more.
Safety rules have moved alongside the technology. A comparative review of the 2011 and 2025 editions of ISO 10218, the core standard for industrial robots, finds that the new edition brings the collaborative-robot specification ISO/TS 15066 into its normative text, introduces new classifications for collaborative applications and adds cybersecurity requirements. The practical message is that collaboration is assessed as a whole application, including tool, workpiece and surroundings, rather than claimed for an arm alone. In welding, where the torch, the arc and the fumes carry their own hazards, that distinction is concrete.
The largest change is in how robots are told what to do. The teach pendant, with its waypoints and parameter fields, is giving ground to methods that start from data a shop already holds. An IFR case study describes an Italian sheet-metal fabricator whose software reads a part's 3D CAD model, identifies the weld joints and generates the robot's positions and transitions automatically, with the program checked offline for collisions before it reaches the cell. Vision systems that locate seams on the part, and research into instructing robots in plain language, push in the same direction: less specifying of motion, more describing of the result.
For operators, that shifts the work from authoring a program to judging one. When software proposes a path, the person at the cell needs to see clearly what the robot intends to do, in what order and why, before the arc is struck, and to correct a single step without starting again. Previews, simulation and plain explanations of automatically generated programs therefore matter more than ever. The interfaces that earn trust on a shop floor will be those that make a machine's intentions legible at a glance, to seasoned tradespeople and newcomers alike.

Education Sharing Platform Web Site Build

Companies with something to teach, students with time to give, and young people who need both — on one platform.
NGOs and businesses offering education, university volunteers, and the schools and teenagers on the receiving end. Each arrives expecting something different.
The structure has to connect roles that do not want the same thing from it. Put the participants and the teaching material on the two axes and the rest follows from where they meet.
We designed a service for each role around those axes and built the site.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, platforms that connect schools with outside educators, including companies, non-profit organisations, university volunteers and working professionals, have become a recognised part of education systems rather than a goodwill extra. Korea built one of the most structured versions. Once the free semester became a standard part of middle school in 2016, schools needed a steady supply of career-experience activities. The government-backed Ggumgil career-experience network lets teachers search for programmes by map or by filters, lets providers manage school applications, and runs a certification scheme for organisations that offer career experience as an education donation.
Elsewhere the same idea has taken other forms. In the UK, the charity Education and Employers runs Inspiring the Future, which connects schools directly with volunteers from the world of work, from primary age upwards. Its research supplies much of the case for doing so. The Drawing the Future survey, published in 2018, asked children aged 7 to 11 to draw the job they wanted. More than a third based their ambitions on people they knew, gender stereotyping in job choices appeared early, and fewer than one per cent had learned about a job from someone who visited their school.
Volunteering infrastructure has matured in parallel. Korea's national 1365 volunteering portal lists opportunities and issues certificates of volunteer activity, which matter to students for whom documented service carries weight. Matching services of this kind tend to find that listing opportunities is the easy part. Keeping supply and demand aligned is harder, since provision is often uneven between cities and smaller towns, and online sessions, which became normal during the pandemic, now serve as a partial answer to distance.
Public platforms also carry obligations that commercial ones can sometimes postpone. Accessibility is the clearest. The W3C's WCAG 2.2 is the current reference for accessible web content, adding success criteria on keyboard focus, target size, alternatives to dragging, redundant entry and accessible authentication, and Korea's national web accessibility standard follows the same principles. For a service used by teenagers, teachers and volunteers alike, those requirements shape sign-up flows, forms and scheduling as much as visual design.
For the people using such platforms, whether a teacher trying to fill a timetable, a volunteer with a free afternoon or a student choosing between unfamiliar options, the experience depends on trust and clarity more than on features. Each needs to see quickly what is on offer, from whom, and what is expected of them, and to feel that arrangements are safe, especially where adults from outside a school work with young people. Platforms that make expectations and safeguarding explicit, and that take the administrative effort out of arranging a session, are the ones that keep participants coming back.

ST Service UI/GUI/VI Design and Guide

A television that tunes itself for a game, and keeps an eye on a child or an elderly parent while it does.
The service linked a smart TV to a phone, and it was asked to do two unrelated things at once.
Nobody pauses a game to adjust the lights. So one question was how the house gets operated without leaving what you are doing; the other was how a carer learns what is happening while it is happening.
We made the interface, the graphics and the visual identity with a guide to go with them. The scope grew during the work.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the television has quietly become one of the most capable computers in the home. It has a large screen, a constant power supply and a place in the room where people gather, and manufacturers have spent the period giving it jobs well beyond showing programmes: games platform, smart-home controller and, more recently, part of how families look after one another.
Gaming changed the set's relationship with the devices plugged into it. HDMI 2.1 introduced Auto Low Latency Mode, which lets a console or PC signal the display to switch into a low-lag mode automatically, and back again when a film starts, sparing players a trip through the picture settings. Cloud gaming then removed the console altogether for some players: Samsung's Gaming Hub brings streaming services including Xbox and NVIDIA GeForce NOW to its 2022 and later smart TVs. The screen increasingly configures itself around what it is being used for.
At the same time, the TV took on a role in the connected home. Recent Samsung sets include a built-in SmartThings hub, so the screen in the living room can act as the controller for lights, sensors and locks, including Matter devices from other brands. Apple TV plays a similar part in Apple's Home ecosystem. The household's largest display has become a natural place to see, at a glance, what the rest of the home is doing.
The newest development is care. Populations are ageing quickly: the World Health Organization expects the number of people aged 60 and over to double to 2.1 billion by 2050, and many families are turning to connected devices to keep an eye on relatives who live alone. Samsung's SmartThings Family Care lets carers check on a relative's well-being remotely, set up routines that flag unusual changes in daily activity, and send medication and appointment reminders, with Samsung TVs from 2024 onwards and refrigerators extending what it can do. Household appliances are being enlisted into a task once handled by phone calls and visits.
For the people involved, the TV's widening role raises distinct interaction questions. Players benefit from settings that follow the content automatically; residents need home controls reachable from the sofa without digging through menus; and in care, both the person being looked after and the carer need to understand what is being observed, who can see it and when an alert is justified. The more a shared screen does on its own, the more its design has to make those behaviours visible, and open to consent, for everyone in the room.

International Design Award Jury Brochure

A hall the size of a gymnasium, thousands of entries, and a few seconds each.
iF and Red Dot jurors move through the room on their feet. A brochure gets picked up, turned, put down, and then it is the next one's turn.
At that speed the layout is the argument. What the product is has to arrive before the juror has decided how long to give it.
Six Galaxy products, printed on a stock chosen to survive a day of being handled. We designed it and saw it through printing and binding.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the large international design awards have been asked to judge more kinds of work, by more explicit standards, under conditions that have not always been kind. iF and Red Dot built their reputations on expert juries examining real objects in a hall, and both still receive thousands of entries a year from manufacturers that treat a win as a mark of credibility.
The pandemic tested that model early. Red Dot's 2020 product design jury, due to gather in Essen in early March, completed its work with a few changes just before Germany's first lockdown; that summer its brands and communication design jury evaluated every entry online for the first time. The organisers reported that the virtual format worked well, yet said they hoped it would remain the exception. Screen-based work translates readily to remote review; a chair, a kettle or a television is harder to judge from photographs, and hands-on examination has kept its place in the leading product awards.
The criteria have shifted too. For its 2025 edition iF replaced its former "Impact" criterion with sustainability, which now stands beside idea, form, function and differentiation as one of five criteria, supported by optional questions and accreditations through which entrants can document social and environmental claims. Jurors are being asked to weigh evidence that cannot be seen on a plinth: materials, repairability, packaging and the life of a product after sale.
The objects themselves have changed. Interfaces, services and whole experiences now compete in categories of their own, and a phone or an appliance is judged as much for its software as for its housing. Samsung Electronics, a regular winner, shows the breadth: its 77 iF awards in 2026 spanned nine disciplines, including UX and UI, recognising its One UI 7 software and a smart-home feature that analyses appliance status and usage patterns alongside televisions, audio products and packaging. For a jury, that means reading behaviour and flows, not only proportion and finish.
For the people who judge, specify and use designed products, these shifts move evaluation beyond appearance. Experts increasingly have to assess claims about software behaviour, accessibility and environmental effect that only reveal themselves in use, which makes clear evidence and comparable presentation part of good design communication. For everyday users, an award label becomes a signal about more than looks, and it is worth something only if the qualities it rewards, such as coherent interaction, durability and ease of use, are the ones people meet once the product is in their hands.

Book Club Integrated Usability Improvement Design

Hundreds of thousands of books, three ways to get any one of them, and a reader who mostly wants to know which one comes next.
Woongjin Book Club opens the company's library to people who are not Woongjin ThinkBig subscribers. The same title can be bought, borrowed, or taken as part of a subscription, and several hundred thousand of them sit behind that choice. At that size the classification is not a convenience; it is the only way anyone arrives at a book at all. So the structure that was already in place was the first thing to take apart.
Buying, lending and subscribing are three different relationships with one object. Treat them as three product lines and the reader is left to reconcile them; hide the difference and it goes wrong exactly where it matters — what is kept, what goes back, and what stops when the paying stops.
Children on the service also choose a reading teacher. That is not the same kind of choice as choosing a book and it cannot be laid out as though it were: the question is trust rather than features, and the person deciding is often the parent rather than the reader.
Usability improvement across the web and tablet interfaces, including the screens where a reader keeps track of what they have.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, digital reading has settled into three overlapping models: buying an e-book, borrowing one from a library, and subscribing to a catalogue. Each implies a different relationship with the text, and the boundaries between them have become a matter of public argument as well as of interface design.
Ownership has proved the least stable of the three. Most e-book "purchases" are in fact licences to read on a particular platform, and that has become harder to ignore. In February 2025 Amazon withdrew the option to download purchased Kindle books to a computer and copy them to an e-reader over USB, leaving Wi-Fi delivery as the route to its devices; coverage noted that the change made personal backups harder and highlighted how little digital content readers actually own. In California, a law in force since 2025 requires sellers of digital goods to make clear when a purchase is really a licence.
Lending moved the other way, becoming easier to use even as its legal basis was tested. Library apps such as OverDrive's Libby let readers borrow e-books, audiobooks and magazines from their local library and download them for offline reading. At the same time, the courts narrowed what libraries can do on their own initiative. In Hachette v. Internet Archive, the Second Circuit upheld a ruling against the Archive's scan-and-lend model in September 2024, and the Archive said that December it would not pursue the case further. Library e-lending therefore continues to rest on licences negotiated with publishers.
Subscriptions became the third pillar, with all-you-can-read services in many markets, Korea among them, and audiobooks increasingly bundled into broader media subscriptions. For children's reading the stakes are not only commercial. The UK's National Literacy Trust reported in 2025 that the share of children and young people who say they enjoy reading had fallen to its lowest level in 20 years, at roughly one in three. Reading services for young people are increasingly judged on whether they sustain the habit of reading, not on the size of the catalogue behind them.
For readers, the trend shifts trust from the object to the platform. People build a digital library over years, across devices and accounts, and they notice when a feature they depended on is withdrawn. Plain statements of licence terms at the point of sale, the ability to export notes, highlights and reading history, and continuity when a device or service changes are becoming part of the reading experience itself, as much as typography or page turns.

Premium Navigation UI Specification

A new generation of the system. The driver's second of attention has not changed; almost everything else has.
More functions than the last generation, and export markets with their own road rules, languages and units.
A generation change is mostly a decision about what not to redo. What gets defined afresh, what is inherited, and where the regional differences are separated out so they do not spread through the whole specification.
We set the common structure first and managed the regional differences as derivations from it. Several people shared the work, so the specification's own conventions were agreed before any of it was written. Standard and premium annual programmes followed.
2 separate engagements on this line, 2017–2019.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the car dashboard has been through a full cycle. Carmakers spent the late 2010s and early 2020s replacing rows of buttons with ever larger touchscreens, some stretching across most of the dashboard. Screens were easier to change than physical tooling, could be revised through software updates and looked modern in a showroom. Climate, seat heating and, in some cars, functions used while driving moved into menus.
The counter-movement has now been written into safety ratings. Euro NCAP's 2026 protocols assess the placement, clarity and ease of use of essential controls, including physical buttons for commonly used functions, which consumer feedback suggests can reduce distraction. In October 2026 it published the first results under the new rules: indicators, wipers and washers, hazard lights, the horn and the eCall button are expected to work through a direct physical input rather than a touchscreen menu. The Audi Q7 scored 90% for driving controls and the Kia EV2 was praised for its controls, while the XPENG X9 managed 20% for its primary driving controls because it relies heavily on the central touchscreen.
The result is a hybrid cockpit rather than a return to the past. Hyundai's Pleos Connect, as reviewed by Jalopnik in September 2026, pairs a large centre touchscreen, a slim driver display and a head-up display with a row of physical toggles for climate, seat heating and ventilation and hazard lights, plus knobs for volume and tuning. The left third of the centre screen always shows driving information. The reviewer found the switches easy to use without looking away from the road, and the menus heavy on swiping.
For anyone planning a new generation of in-car software, the question has changed. It is less about how much can be placed on the screen and more about which functions deserve a fixed physical place, which deserve a fixed zone on the display, and which can live in menus. Information is now deliberately repeated across the driver display, the head-up display and the centre screen. Software can keep changing after sale while switches cannot, so the split between them has to be right at launch. Although Euro NCAP is a European programme, cars are increasingly developed for several markets at once, so its preferences are likely to shape cabins well beyond Europe.
For drivers, the lesson of the last decade is that predictability beats flexibility for the things done most often and in a hurry. A control that stays in the same place can be found by touch and memory; a menu item that moves with an update cannot. The best current cabins treat the touchscreen as the place for setting things up and browsing, and keep the actions a driver needs without warning within reach of a hand that never has to look.

Learning Model Platform Service Design

Children already tired of homework had to want to open it again tomorrow.
An AI learning platform for primary school children on a tablet, with parents watching the results.
Words like revision and preview belong to adults. So the plan was written in the language children actually use — yesterday, tomorrow — and we had to decide how much of a child's progress a parent should be shown.
Achievement and reward were built into the structure rather than added on top.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, AI in children's learning has gone through two distinct phases. The first was adaptive learning: platforms that adjusted the difficulty and order of exercises according to a child's answers, usually delivered on a tablet with a companion app for parents. Korea's large home-learning companies moved early here, alongside a global wave of adaptive maths and literacy products. The second phase, from 2023, brought generative AI into the picture, and with it a different kind of interaction.
Khan Academy's Khanmigo is the best-known example of the newer approach: an AI tutor and teaching assistant designed to guide learners towards answers through questions rather than simply supplying them. The design choice responds to an obvious risk, that a capable chatbot will do homework instead of supporting it, and it has become a reference point for education products built on large language models.
Policy followed quickly. In September 2023 UNESCO published its first global guidance on generative AI in education and research, calling for protection of learners' data privacy, age-appropriate use guided by humans, and an age limit for children using such tools on their own. In the UK, the Information Commissioner's Age Appropriate Design Code sets fifteen standards for online services likely to be used by children, among them high-privacy default settings, limits on nudge techniques that push children to give up data or weaken their privacy, and a standard on parental controls. These frameworks bear directly on features that learning platforms depend on: profiling to personalise content, engagement mechanics, and monitoring tools for adults.
Design research with children has also moved on. Nielsen Norman Group's 2019 study of users aged 3 to 12 found that children had become considerably more proficient with websites and apps than in its earlier studies, while many designs still failed to address narrow age bands, since what suits a six-year-old rarely suits a ten-year-old. Gamification in the form of points, badges and streaks remains common, but its value depends heavily on how it is designed and where it is used, and reward schemes that rely only on external incentives can lose their pull over time.
For children, interaction is shifting from tapping through exercises to talking with a system that answers back. That makes tone, honesty about what the system is, and limits a young user can understand more important than before. For parents and teachers the central question becomes one of oversight: being able to see what an AI tutor said and why, not only the scores that came out of it. Products that keep learning in the child's hands while making adult oversight visible and understandable to the child are likely to earn the most trust.

Link Emulator UX Consulting

A link emulator is a test instrument that simulates tactical data link networks such as Link 16 and Link 22, so that data link equipment can be developed, tested and analysed without live participants.
A link emulator reproduces the participants and message traffic of a tactical data link network, such as Link 16 or Link 22, in a laboratory or test environment. Instruments of this kind are used to develop and integrate a platform's data link equipment, to check that its messages conform to the link standards, and to record and analyse network traffic, reducing the need for live units in testing.
AEON provided UX consulting for a link emulator used for tactical data link emulation and analysis.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Tactical data links are how ships, aircraft and command posts share a common picture in near real time. NATO navies have long relied on Link 11 and Link 16; Link 22, a beyond-line-of-sight successor to Link 11, is now entering service, while the Joint Range Extension Applications Protocol (JREAP) carries J-series messages over satellite and IP networks. Few platforms run only one of these, and the complexity lies in making them work together.
Link 22 adoption has gathered pace. During the Clemenceau 25 deployment, the French Navy tested Link 22 with the Charles de Gaulle carrier strike group, exchanging data with US and Japanese ships in exercise Pacific Steller, with the aim of eventually replacing Link 11. In July 2026 USS Mount Whitney became the first US warship to integrate and operate Link 22 organically while underway, connecting with the German frigate Hamburg during BALTOPS 2026.
Terminals are changing too. MIDS JTRS, a four-channel software-defined radio that runs the Link 16 waveform alongside up to three other protocols, is being supplied for more than 45 US and international platform types, with naval vessels among the platforms on which it is fielded. Software-defined terminals make it easier to add waveforms and cryptographic updates, but they also multiply the configurations that must be proved interoperable.
That is why testing and emulation have become an industry of their own. Tools such as Curtiss-Wright's TCG BOSS script and generate messages, simulate networks, emulate terminals in real time and analyse results afterwards, and are used to certify that data link implementations conform to military standards; they can also populate a test with virtual command posts, weapons and threats. Sea trials remain essential: in 2013 Finland's acceptance testing of its multi-link data link processor exercised Link 22, MIDS/Link 16 and JREAP together aboard the minelayer Uusimaa, and trials of this kind have become routine as navies integrate several links through a single processor.
The direction is towards multi-link gateways, data forwarding between networks, dynamic network management and crypto modernisation, with link managers coordinating a growing web of participants across services and nations. The more links a force uses, the more it depends on laboratories and emulators that can reproduce realistic network traffic before equipment goes to sea.
For operators, the hardest part is often invisible. Link managers and interface control officers must understand the state of several networks at once, notice when tracks are duplicated or dropped, and diagnose faults that may lie in a terminal, a gateway or a message translation. Test and monitoring tools that make network behaviour legible, rather than burying it in message logs, help engineers in the laboratory and operators at sea alike.

Samsung -ONE UI Company-wide Rollout for Reference Design

One UI was drawn for a phone. It had to reach a watch, a television and a refrigerator.
The same design language on screens of very different sizes, looked at from very different distances.
What travels unchanged is what makes it one language; what has to change is what keeps it usable. Separating those two is the whole of the rollout.
We analysed how each device is actually used, built reference designs from that, and folded in what came back from an online survey.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s the major platform owners have stopped treating their design languages as skins for a phone and started treating them as identities that must hold across everything they sell. Samsung's One UI, unveiled at its developer conference in November 2018, was an early example. Built around one-handed use of large phones, it has since spread to tablets, foldables, watches and laptops, and Samsung has more recently carried the name onto its Tizen-based televisions and home appliances.
Rivals have followed the same path in their own ways. Google's Material You introduced colour drawn from the user's wallpaper in 2021, and Material 3 Expressive, announced in May 2025, extended that system across Android 16 and Wear OS 6 with springier motion, more emphatic type and a watch design built around round displays. In June 2025 Apple introduced Liquid Glass, a translucent material applied across iOS, iPadOS, macOS, watchOS and tvOS in a single release.
The reach of these systems has raised the cost of getting them wrong. Nielsen Norman Group's review of iOS 26 argued that Liquid Glass weakened legibility and discoverability, citing translucent controls over busy content, unlabelled back buttons and controls that appear and vanish depending on context. When one design language spans hundreds of millions of devices, a choice that looks refined on a stage becomes a daily readability problem, and it cannot be quietly withdrawn.
The way such systems are built has changed as well. Design languages are increasingly maintained as libraries of tokens and components that generate variants for each kind of device, rather than as a set of reference screens adapted by hand for every product team. Personal colour, motion that responds to touch and type that adapts to the display are now expected rather than remarkable, and AI assistants are beginning to sit across all of it as a common layer.
For the people using these devices, consistency is worth exactly as much effort as it saves them. A shared visual language helps when habits learned on a phone carry over to a watch or a television without relearning; it gets in the way when brand expression is put ahead of contrast, reach or readability. The real test of a cross-device design language is whether it recedes behind the task on every screen it touches.

New Wearable Project UI Development

Two things worn at once, and neither of them is the phone. The earbuds have no screen, so the watch becomes the only place they can be seen.
Galaxy Buds2 and Galaxy Gear are both on the body, and the work was what passes between them. Everything about the earbuds a person might need to see has to appear somewhere other than the earbuds — which of them is connected, how much charge is left, what mode they are in. The nearest surface is the wrist.
A wrist screen is read in glances of about a second, so it cannot simply become a small second screen for the buds. What earns a place there has to be weighed against what could be a tone in the ear instead, and against what should wait for a screen large enough to deserve it.
The two channels are not interchangeable, because one of them is public. Raising a wrist is something other people see; a sound in the ear is not. Which of the two carries a given message is a decision about more than convenience.
UI development for the pairing, run in two phases across more than a year.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the earbud has turned from an accessory into a small computer in its own right. When Samsung's first Galaxy Buds arrived in early 2019, the company's watches still ran Tizen, and true wireless earbuds were judged mostly on fit, battery life and sound. Today's premium models carry their own processors, sensors and on-device machine learning, and the interesting question is no longer how to pair two gadgets but how several devices worn on one body should divide the work between them.
Platform consolidation set part of the stage. In 2021 Google and Samsung announced a unified wearable platform that folded Tizen's strengths into what became Wear OS 3, giving developers a single target and moving Samsung's watches onto the same software base as much of the wider Android watch market. Coordination between a watch, a phone and a pair of earbuds became something the platform was expected to handle, rather than something each manufacturer improvised.
Input, meanwhile, has been moving off the touchscreen. Apple's double tap gesture, which reached Apple Watch Series 9 and Ultra 2 in October 2023, lets the wearer pinch finger and thumb together twice to trigger the main action on screen, such as answering a call or snoozing an alarm, without touching the display. Samsung has since added a comparable double pinch to its own watch software. The following year Apple let AirPods Pro users nod or shake their heads to respond to Siri announcements, pitching it as a way to reply without speaking aloud. In both cases the body becomes the controller, and the screen becomes optional for the most frequent responses.
The most consequential change, though, is clinical. In September 2024 the US Food and Drug Administration authorised the first over-the-counter hearing aid software, allowing compatible AirPods Pro to act as a self-fitted hearing aid for adults with perceived mild to moderate hearing loss. A product bought for music and calls now also sits inside medical-device regulation, with the fitting carried out by its owner through a phone rather than by an audiologist. Features that once required dedicated equipment are increasingly arriving as a firmware update.
For the people wearing these devices, the practical result is that information and control are spread across several small surfaces, each tied to a different sense: sight on the wrist, sound in the ear, movement of the head and hand. Good interaction design here depends less on any single screen than on coherence across them. The same gesture should mean the same thing wherever it is made, confirmation should come through a channel the wearer is already attending to, and anything that touches health deserves clearer explanation than a control that skips a track.

Utilizing Satellite Image Data Planning Projects

This 2018 planning project examined how LIG Nex1 could work with Planet Labs, the US operator of an Earth-observation satellite constellation, to develop civilian demand for satellite imagery.
Planet Labs, based in San Francisco, operates a large constellation of small Earth-observation satellites that image the Earth's land surface at high frequency, and sells the data and analytics to government and commercial customers. Civilian applications for such imagery include agriculture, forestry, mapping, infrastructure monitoring, insurance and disaster response, alongside defence and intelligence uses.
AEON built the future strategy and business case for LIG Nex1 (now LIG D&A) on working with Planet Labs to develop civilian demand for satellite imagery, and presented them.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Commercial Earth observation has become a core defence capability. Since Russia's full-scale invasion of Ukraine in 2022, imagery from private operators has been used openly by governments and armed forces, and the US National Reconnaissance Office signed long-term commercial imagery contracts with Maxar, Planet and BlackSky under its Electro-Optical Commercial Layer programme. The line between civil and military customers, once sharp, is now blurred.
Radar satellites have gained much from this shift. Synthetic aperture radar sees at night and through cloud, and Ukraine showed how a buyer could secure priority: in 2022 it bought exclusive access to an ICEYE satellite with crowd-funded money, pushing the industry towards dedicated-satellite offerings. Planet, best known for fleets of small optical satellites, has moved towards higher resolution with its Pelican series, and in 2025 signed a multi-year agreement giving the German government dedicated capacity and direct-downlink access to Pelican.
South Korea has built a sovereign layer alongside this market. The 425 Project was completed with the launch of its fifth satellite in November 2025, giving the military one electro-optical/infrared and four SAR satellites of its own. The next step is numbers: the military's plan calls for around 60 small and micro reconnaissance satellites by 2030, aiming to watch the peninsula at intervals of 30 minutes or less. On the civil side, the NEONSAT microsatellites launched on the Nuri rocket in October 2026 sent back their first images within a day of launch.
The business logic now runs in both directions. Defence demand provides anchor revenue that lets companies build and refresh constellations, while civil demand from agriculture, insurance, infrastructure and disaster response keeps unit costs down and revisit rates high. Value is moving towards analytics: customers increasingly buy answers, such as detected changes or counted objects, rather than raw images, and governments mix sovereign, allied and commercial sources in a single collection plan.
For analysts and operators, the constraint is no longer the number of images but the attention available to examine them. Frequent revisits from many satellites produce more data than people can inspect, so automated change detection and object recognition increasingly decide what reaches a human. Interfaces need to show what was imaged, when, by which sensor and with what confidence, and make comparison across time effortless. Tasking is changing too, from ordering a single image to managing a standing watch across sovereign and commercial sources.

One Magician UX Project UX Prototype and Film

A table of figures tells you what the drive is doing. It does not tell you whether that is good.
Samsung's memory division ships a great many products, and this is the single piece of software that manages all of them. Whatever it shows has to hold across parts that differ in what they are even able to report about themselves.
What was there was a screen of numbers. A number is a measurement, and the person reading it wants a judgement — someone who has just bought a drive does not know what a given transfer rate means, or whether a given temperature is a problem. They open the software once, to find out whether the thing is all right.
So the work was to put the interpretation on screen beside the measurement and let graphics carry it. That is not decoration — it is the software answering the question the person actually arrived with. The figures stay for anyone who wants them.
UX prototype for the next version, and the film that explains it.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, consumer storage has changed faster than the software people use to look after it. In 2018 a typical upgrade meant swapping a hard disk for a SATA solid-state drive; today mainstream machines ship with NVMe drives on PCIe 4.0, and PCIe 5.0 parts are moving into enthusiast desktops. Each step has brought more internal telemetry, from wear levels and spare capacity to error counts and temperature, and more pressure to make it intelligible to people who never asked to think about it.
One clear shift is that the operating system has begun to take on part of this job. In 2020 Microsoft started testing storage health monitoring for NVMe drives in Windows 10, raising a notification when a drive shows signs of failure and pointing users to a drive health section in Settings with the warning and further details. Health reporting that once lived only in manufacturers' utilities and enthusiast tools has become a platform feature, and its central instruction is a plain one: back up now.
Large fleets have also made reliability something the public can inspect. Backblaze, which publishes statistics from its own data centres, has reported lifetime annualised failure rates for its SSD boot drives just under one per cent, while noting that its SSDs are at an earlier stage of life than its hard drives and that some models' figures carry wide confidence intervals. Anyone presenting drive health to consumers inherits the same problem: a single figure carries more uncertainty than it appears to.
That point was made sharply in early 2023, when owners of Samsung's 990 Pro reported health readings falling far faster than expected. Samsung traced the fault to firmware and issued a fix, most easily applied through its Magician utility; as Puget Systems noted, the update stopped the accelerated decline but did not restore readings already lost. The episode showed that a health indicator is itself software, and that the tool which displays it is also the route by which it gets corrected.
Heat is the newer complication. Early PCIe 5.0 drives drew enough power that reviewers found some could throttle, raise file-system errors or even shut a PC down without a heatsink, and controller makers responded by revising their thermal throttling behaviour while drive makers added integrated coolers. Temperature has moved from a line in a specification sheet to a factor that decides whether a drive delivers the speed printed on its box.
For users, storage has become something the computer increasingly speaks up about, rather than something checked only after it fails. The design challenge lies in pacing those interventions: warnings that are rare, specific and tied to a clear next step, such as making a backup or applying an update, and enough openness about how health is estimated that people can calibrate their trust instead of treating a percentage as a verdict.

Open IoT Keynote Film and UX

A watch on an operating system of its own has very few apps. The way out was not to write more of them.
Galaxy Watch ran Tizen, and Tizen could not borrow from anywhere else. The shortage of applications was not a backlog; it was the platform.
If the watch cannot host the app, it can still be the thing that reaches it — which turns a platform problem into a connected-device one.
We set a reference brand to make it concrete, showed the linking scenarios as film for the developer keynote, and designed the interface and the guidelines behind them.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the smartwatch market has consolidated and its software has changed shape. In 2018 several watch platforms competed for developers' attention, each with its own tools and store. That picture simplified in May 2021, when Google and Samsung announced that Wear OS and Tizen would be combined into a single platform, promising faster app launches, better battery life and one target for developers instead of two. Samsung's watches moved onto the shared platform later that year.
At the same time, the way people use a watch has drifted away from the app grid. Tiles, complications and quick panels now carry much of the everyday interaction, and a growing share of it concerns devices elsewhere in the home. In early 2023 Samsung extended SmartThings on Galaxy Watch to thermostats, air purifiers and blinds alongside TVs and lights, added live views from Ring and Nest cameras, and placed the controls a single swipe from the watch face. Google took a similar route in 2024, adding a favourites tile, watch-face shortcuts and finer device controls to Google Home on Wear OS.
Interoperability made this more practical. Matter 1.0, released by the Connectivity Standards Alliance in October 2022 with contributions from Amazon, Apple, Google, Samsung SmartThings and hundreds of other members, gave lights, locks, blinds, sensors and media devices from different manufacturers a common language. A small screen on the wrist is far more useful as a controller when the things it controls no longer depend on one brand's ecosystem.
The newest layer is conversational. In May 2025 Google said that Gemini would come to Wear OS watches that already ran its assistant, alongside a Wear OS 6 redesign built around the round display. Natural-language requests reduce the need to navigate menus on a screen a few centimetres across, although they still depend on a network connection and on the assistant understanding which device, room or task the user means.
Taken together, these changes alter what a watch interface is made of. The open questions are less about how many apps a platform offers and more about trust: whether a command reached the right device, whether a camera feed is current, whether the watch shows the true state of a door or a thermostat.
For the people wearing these devices, interactions last seconds, attention is divided and the thing being controlled is often out of sight. That puts weight on clear confirmation of outcomes, controls that sit in predictable places, honest handling of failure, and voice and touch paths that lead to the same result.

Samsung Card Open Sharing Season20 Social Goods Project

Young people with an idea, and no route from the idea to a shelf.
The programme took social-goods ideas through to products actually on sale.
Most of that is not design; it is the parts nobody plans for — finding mentors worth the young teams' time, running the talks and workshops, getting mock-ups made, and walking each team through what it takes to get listed in a shop.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the idea that a small enterprise can sell an ordinary product while pursuing a social purpose has moved from the edge of policy towards its centre. Korea was early. Its Social Enterprise Promotion Act came into force in 2007 and created a government certification route: the labour minister decides, after review by a support committee, which organisations qualify, and their status is published. Around that framework grew a layer of incubators, foundation programmes and corporate partnerships that try to turn a good idea into something that can be manufactured and sold.
International bodies have since written the same ambition into their own instruments. The OECD's Recommendation on the Social and Solidarity Economy and Social Innovation, adopted in 2022, asks governments to open public and private markets to these organisations, including through online marketplaces, and to encourage them to measure and report their impact. In April 2023 the UN General Assembly adopted a resolution on the social and solidarity economy, connecting it to decent work, social inclusion and the Sustainable Development Goals, and drawing attention to access to finance and markets.
The practical route from prototype to shelf has also shortened. Crowdfunding campaigns, marketplace storefronts and short retail placements let a young team test demand before committing to a full production run, and a product can find its first buyers without a distributor. What has not become easier is being believed. Shoppers meet a steady stream of goods that say they do good, and most have no way to check.
Regulators have started to address that gap directly. The EU's Directive (EU) 2024/825 adds a product's environmental or social characteristics to the things a trader can mislead consumers about, and its recitals give working conditions and contributions to social initiatives as examples of social traits. It also bans sustainability labels that rest neither on a certification scheme nor on a public authority. The rules have applied since 27 September 2026. The direction of travel is visible well beyond Europe: a social claim on a product is increasingly treated as a factual statement that can be tested, not as a tone of voice.
For the people who make and buy such goods, this changes what a product page or a package has to do. A story still draws attention, but it now has to sit next to something checkable: who certified the claim, what was measured, where the evidence lives. Interfaces that present impact need to make that evidence reachable in a step or two without burying the product itself, and to keep the visual difference between an audited label and a brand's own badge obvious. For small teams, the skill that matters is turning honest, limited facts into something a hurried shopper can read in a few seconds.

Warrior Platform BLOCK-I UX Consulting

Warrior Platform is the Republic of Korea Army's soldier modernisation programme, covering individual equipment, wearable devices and the systems they report into.
The programme aims to equip each soldier to function as an independent combat platform, with items ranging from helmets, body armour and combat uniforms to sights and communications equipment. It is being introduced in phases as part of the Army's wider modernisation of its combat units, with Block-I as the initial increment.
AEON provided hardware and software UX consulting for Warrior Platform Block-I.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Almost every major army has tried to turn the infantry soldier into a networked node, and the history of these programmes is a lesson in how hard it is to put computing on a human body. Weight, power, heat and cognitive load all limit what a soldier can carry and actually use under fire, and each new device competes with the rifle, the radio and the soldier's own eyes for attention.
The US Army's experience is the most public. The Integrated Visual Augmentation System, a HoloLens-derived headset, was meant to fuse night vision, maps and targeting into a single heads-up display. According to the Congressional Research Service, users reported headaches, eyestrain, nausea and poor low-light performance, and in July 2026 the GAO reported that none of the roughly 10,000 units developed would be fielded. The successor, Soldier Borne Mission Command, is being prototyped by Anduril and Rivet; Anduril says it has built manufacturing capacity for thousands of systems, with a production decision possible in fiscal 2028.
European programmes have taken a more incremental route. France's FELIN combined optronics, compact radios and an information terminal around the soldier and his weapon. Germany is digitising its infantry through IdZ-ES: a 2025 framework contract with Rheinmetall upgrades existing platoon systems to a digital standard able to exchange data with Boxer and Puma vehicles. Buying the platoon, rather than the individual, as the unit of procurement reflects a view of the soldier as one element of a larger networked system.
South Korea's Warrior Platform follows a similar philosophy. The Army describes it as equipping each soldier to act as an independent combat platform, with items from helmets and body armour to sights and communications equipment introduced in phases. The newest element across all these programmes is the small drone: unmanned aircraft and their video feeds are becoming standard squad equipment, and the soldier's display increasingly has to show what the drone sees as well as where friendly forces are.
The lesson of two decades of soldier modernisation is that the human is the limiting component. Displays that work in a laboratory can fail at night, in rain, with gloves on and while running. Information that helps a platoon commander can overwhelm a rifleman. The systems that succeed tend to deliver less information but the right information at the right moment, tolerate being ignored when the soldier is busy, and are tested with troops early enough for their feedback to change the design rather than merely to document its problems.

S-PDU(Pilot Display Unit) Prototype UX

A pilot display unit is a cockpit display that presents flight, navigation and mission information to the pilot.
Pilot display units range from fixed instrument-panel screens to portable and tablet-based displays, which can carry the maps, charts, checklists and mission data that once filled paper flight bags. Portable and tablet-based cockpit displays are increasingly built on commercial computing hardware and operating systems, which allows new functions to be added through software.
AEON carried out the software and hardware UX for a pilot display unit and built a prototype on an Android handset.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The cockpit has spent decades moving from dials to glass, and the latest generation of combat aircraft has pushed that logic to its limit. The F-35 replaced the traditional cluster of screens with a single 20-by-8-inch panoramic display that can be operated by touch, cursor or voice, and dispensed with the fixed head-up display altogether, projecting symbology onto the pilot's visor instead.
The helmet has become a primary instrument. Collins Elbit Vision Systems has delivered more than 3,000 Gen III helmets for the F-35, whose interface the company describes as the pilot's primary display system. Its successor, the Zero-G HMDS+, passed critical design review for the US Navy in December 2025 and is planned for fielding on Super Hornets and Growlers, with initial capability expected in 2027, with lighter weight intended to reduce neck strain and improve ejection safety. The F-35's own panoramic display is being renewed under the Technology Refresh 3 programme, with Elbit Systems of America developing the replacement display unit.
In parallel, the portable display has gone mainstream. Electronic flight bags, usually commercial tablets, have replaced paper charts and manuals across many air arms. In one US Army combat aviation brigade, a single tablet replaced around 20 separate publications, and mission changes could reach every crew within seconds. Tablets bring commercial development speed and familiar touch interaction into military aviation, but they also raise questions of certification, cyber security, mounting, glare and night-vision compatibility.
The direction of travel is convergence. Fixed displays are becoming larger and more software-defined; helmets are adding sensor video and richer imagery; tablets are being connected to aircraft data rather than serving only as electronic paper. The same information may now appear on the panel, in the visor and on a kneeboard device, and the central design question is deciding which information belongs where and when.
For pilots, more display area does not automatically mean better awareness. Each surface competes for attention at moments of high workload, and a symbol that is helpful on a large panel can be distracting in the visor. The trend puts a premium on consistent symbology across devices, on interaction that works with gloves, under g-loads and in turbulence, and on decluttering that adapts to the phase of flight. Testing interfaces with pilots early, in simulators and in flight, remains the surest way to find where attention breaks down before the design is fixed in certified hardware.

Life Care Robot Service for UX and Application Development

The robot lives with an older person. It reports blood pressure and sleep to someone who has never asked a machine for anything.
At home the users are elderly people and their families. Readings — blood pressure, heart rate, sleep — have to land with someone who is not comfortable with devices.
Whether a result is given as a number or said in words is not a matter of presentation; it decides whether it is understood. And a robot that speaks first is a different thing from one that waits, so when it does each had to be settled — along with how much belongs to the robot and how much to the family's remote scheduling app.
We drew on our earlier work on the URC home robots and designed the robot screens and the app flow together, taking the app far enough to be demonstrated. It was shown at CES the following year.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the idea of a robot that keeps an older person company at home has moved from trade-show demonstration to small public programmes, and then to a new generation of devices built on large language models. The pressure behind it is demographic. The World Health Organization projects that one person in six worldwide will be aged 60 or over by 2030, and that the share of over-60s will nearly double between 2015 and 2050. Health systems and families alike are looking for ways to support people who want to go on living in their own homes.
In 2019 the category had its showcase moment. At CES that January, Samsung gave the first public demonstration of Samsung Bot Care, presented as a robot that helps people manage their daily health routines, alongside sibling robots for air quality and retail. Demonstrations like these showed what was possible on a stage. The harder questions, such as reliable sensing in cluttered homes, who is responsible when a reading looks wrong, and what a machine should do when it cannot reach anyone, took longer to answer.
Public deployment came from a different direction: simpler devices focused on companionship rather than measurement. In 2022 the New York State Office for the Aging began delivering ElliQ, a tabletop companion from Intuition Robotics, to the homes of 800 older adults as part of an effort against social isolation. The device was promoted as proactive, opening conversations rather than waiting to be addressed. That stance, once a design experiment, is now treated as a defining feature of the category.
Generative AI has widened what such devices claim to do. In April 2025 Samsung and Google Cloud announced that Ballie, Samsung's rolling home companion, would use Gemini models for natural conversation, reminders and proactive help around the home. Conversation is now relatively easy to build, so the open questions have moved towards consent, privacy in shared households, and how much an always-present machine should volunteer about what it notices. Anything that measures vital signs or offers medical advice falls under medical-device rules, which keeps most companion products on the wellness side of that line, and independent evidence on health outcomes from home robots is still limited.
For older users, the interaction problem is less about making the robot clever than about making it legible: when it will speak, what it is about to do, and what happens to what it hears. For relatives and carers who receive information at a distance, the challenge runs the other way, getting enough to feel reassured without turning someone's home into a monitored space. Designs that leave the older person in charge of what is shared, and that fail gracefully when the technology does, are the ones most likely to be trusted.

Samsung Account UX Development

The same account, signed into from a phone, a television, a refrigerator and a browser. Only one of those has a keyboard anyone would want to use.
Samsung Account is the door to all of them, and every device opens it differently.
One policy, four ways of typing. The flow has to stay recognisable while the input method changes underneath it.
We built prototypes that ran on mobile, TV, Family Hub and the web, and used them to show what the improvements would be.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, signing in has become one of the most cross-device problems in consumer technology. One account now opens a phone, a television, a watch, a car and a home appliance, and few of those offer a keyboard anyone would choose to type on. The direction of travel has been consistent: move the typing to the device that is good at it, and then remove the typing altogether.
The first half of that is now an established pattern. The IETF's OAuth 2.0 Device Authorization Grant (RFC 8628), published in 2019, standardised the approach in which a television or similar device shows a short code and a web address, and the user approves the request on a phone or computer while the device waits. QR codes have since shortened that path further. Samsung's current support guidance, for instance, describes signing a TV into a Samsung account by scanning an on-screen code with a phone that is already signed in, keeping remote-control entry as a fallback.
The second half is the passkey. In May 2022 Apple, Google and Microsoft jointly committed to expanded support for the FIDO standard, including sign-in credentials available across a person's devices and the ability to use a phone to sign in on a nearby device regardless of its operating system or browser. Samsung brought the idea to its own credential manager: from One UI 6, Samsung Pass can create and store passkeys, held in encrypted storage on the Galaxy device.
Four years on, the evidence suggests passkeys are moving from deployment to habit. The FIDO Alliance's Passkey Index 2026, published this month, reports that passkey sign-ins succeed more often than conventional multi-factor sign-ins and that more than a third of successful sign-ins at participating services now use one. The figures are self-reported by member companies, but they match what users see: fewer password fields, more biometric confirmations and more hand-offs between devices.
What remains difficult is coherence. A user may meet a QR code on the television, a fingerprint prompt on the phone, a code on the watch and a password field on an older browser, all for the same account. Each method is sensible on its own; together they can feel like several different services. Recovery is the other weak point. When the phone becomes the key to everything, losing it becomes a much larger problem than it used to be.
For users, the practical shift is from remembering secrets to recognising requests. Interface work moves with it: making it obvious which device is asking and why, keeping the sequence of steps familiar whether it starts on a screen across the room or in the hand, and designing recovery paths that a person under stress can follow without calling for help.

Smart TV Mass-Production GUI Guide

An annual engagement on the mass-production GUI guide for smart televisions and the site that carries it, absorbing specification changes as they arrived through the year.
Finding is the function. The structure that gets someone to the rule they need decides whether the guide is used or worked around. We built it as a website, and the scope grew.
2 separate engagements on this line, 2017–2018.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the interface guideline has turned from a document into a system. At the time, much of the consumer electronics industry still specified its screens through large static volumes, refreshed for each model year and passed between design teams, software partners and factories. That model has steadily given way to what Nielsen Norman Group describes as a design system: a style guide, a component library and a pattern library, kept alive by a standing team rather than issued once and archived.
The most consequential change is that design decisions are increasingly stored as data. Colours, type sizes, spacing and motion values, known as design tokens, can be defined once and turned into code for each platform. In October 2025 the W3C Design Tokens Community Group published the first stable version of its specification, a vendor-neutral format that supports themes and can generate code for iOS, Android and the web. Major design tools have committed to it, which means a rule can travel from a design file to a production build without being retyped along the way.
Platform owners, meanwhile, have shown how far a single design language can be stretched. In June 2025 Apple introduced Liquid Glass, a translucent material applied at once across iOS, iPadOS, macOS, watchOS and tvOS, with updated developer APIs. A month earlier Google announced Material 3 Expressive for Android 16 and Wear OS 6, with springier motion, updated dynamic colour and more emphatic type. Both treat the language as something that must hold from a watch face to a living-room screen, which is close to the problem television makers have always had.
Television brings its own pressures. Interfaces are read from across a room and navigated with a directional remote, so focus states, safe areas and legibility at distance carry the weight that touch targets carry on a phone. Software support periods have also lengthened markedly, with leading manufacturers now promising several years of operating system upgrades for their sets. A guideline therefore has to remain coherent across many model years at once, and versioning, deprecation and backwards compatibility have become as much a part of design governance as the visual rules themselves.
For the people who use these products, the effect of all this machinery should be almost invisible. A viewer notices consistency mainly when it fails: a focus highlight that behaves differently in one app, a menu that moves after an update, text that cannot be read from the sofa. Systems built on shared tokens and components make those lapses rarer, and they make it easier to carry accessibility settings such as larger text and higher contrast through every screen rather than only the ones someone remembered.

Smart Home Accessibility UI Guideline Development

A smart home that only opens for people who can see it, hear it and reach it is not a smart home.
The service had to work for users with limited sight, hearing or movement.
Accessibility standards are written in general, and a house is not general. Contrast, type size and spoken guidance each mean something particular once the thing being controlled is a boiler or a door.
We took the standards and wrote them into the smart home context as UI guidelines. The scope grew during the work, and the contract with it.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the hardest accessibility questions in the home have been about where the interface actually is. Guidance was written for web pages and desktop software, while homes filled up with thermostats, locks, speakers and appliances controlled through a mix of small panels, phone apps and voice. The W3C has spent years adapting its rules to that reality. Its WCAG2ICT note, updated in December 2025 to cover WCAG 2.2, explains how the web criteria apply to non-web software and documents, and gives a section to closed functionality: products such as smart speakers, smart displays, smart TVs and IoT devices that do not let users attach their own assistive technology. It is candid that meeting the criteria alone will not make such software accessible, and it leaves hardware out of scope.
Law has moved, but selectively. The European Accessibility Act sets requirements for products placed on the EU market after 28 June 2025, yet its product list covers computers and their operating systems, payment and other self-service terminals, communications and audiovisual terminal equipment, and e-readers. Household appliances and smart-home devices are not named. Much of the connected home therefore still rests on voluntary practice, even where the phones used to control it fall under binding rules.
The other large change is interoperability. The Connectivity Standards Alliance released Matter 1.0 in October 2022, developed by more than 280 member companies including Amazon, Apple, Google and Samsung SmartThings, so that certified devices could work across brands and platforms. Matter 1.5, announced in November 2025, added cameras, a unified model for closures such as shades, gates and garage doors with more precise position reporting, and energy tariff and EV charging information. Each release widens the set of household functions that any compliant controller can read and operate.
That moves accessibility away from the individual appliance. If a lock, a blind and a thermostat can all be reached from the same phone, the accessibility of the home increasingly depends on the controller a person already uses, with its screen reader, switch access, voice input and display settings. Standardised device models help, because a closure that reports its exact position can be announced in words as well as drawn as an icon. The device itself still matters when the phone is out of reach, and for states that must be perceivable without any screen at all.
For residents with limited sight, hearing or movement, the promise is a home they can run through whichever interface suits them, rather than one learnt appliance by appliance. The risk is that responsibility falls between makers: the device maker assumes the app will cope, and the app assumes the platform will. Good smart-home design now means checking the whole chain, from the physical control to the spoken confirmation, as a single experience.

System Air Conditioner Wired Remote

It is screwed to a wall in an office. Everybody touches it and nobody has read the manual.
A wired remote for a system air conditioner is shared property — in a shop, an office, a corridor.
On a screen that small, the default view is the product. Which of operation, temperature and mode is simply there, and how the administrator's settings are kept away from whoever is passing.
Building on the earlier graphics and visual identity work, we filled in the detail screens and the exceptions that commercialisation asks for.
2 separate engagements on this line, 2016–2017.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the wired wall controller has outlived several predictions of its disappearance. Smartphone apps, voice assistants and cloud dashboards all arrived, yet commercial air-conditioning and heat pump systems still ship with a panel on the wall. It needs no account or network to work, it sits where people already look for it, and building codes and installers rely on it as the dependable local interface. What has changed is the job that panel is asked to do.
The first change is heating. In its 2022 global outlook, the International Energy Agency described heat pumps as the central technology for low-carbon heating, with EU sales up by about 35 per cent in 2021. Equipment once sold mainly for cooling is now expected to carry a building through winter as well, so the controller has to make sense of defrost cycles, auxiliary heat, flow temperatures and schedules that follow electricity tariffs rather than office hours.
The second change is the refrigerant. Europe's revised F-gas rules, adopted in 2024, bar new equipment that relies on fluorinated gases where workable alternatives exist, and set 2035 as the end point for split air conditioning and heat pumps. The industry's interim answer, lower-impact refrigerants such as R-32 and R-454B, is mildly flammable, and the safety standard UL 60335-2-40 requires leak detection above a charge limit, with detection switching on the indoor fan to dilute any escaped refrigerant. A sensor fault, too, keeps the fans running until it is replaced. These are states an occupant may notice without understanding, and the wall panel is often the only place they can be explained.
The third change is that the panel is no longer the only controller. Matter, the smart-home standard, added thermostat presets and scheduling in version 1.4, alongside energy-management device types that include heat pumps. Commercial systems have their own equivalents through building management integrations. In practice a setpoint may be scheduled in an app, overridden at the wall and then nudged by a demand-response signal from the grid, all within an afternoon.
For the people who touch these controllers, the design problem has moved from operating a machine to understanding a system. Someone standing at the wall needs to know why the unit is doing what it is doing: whether a schedule, a remote user, a grid event or a safety function is in charge, and whether their own input will hold. Clear statements of system-initiated states, consistent behaviour across the app and the wall, and safety messages written for people who are not technicians are becoming as important as the temperature figure itself.

One App Plug-ins

One app, many machines, and a different plug-in for each one.
Samsung had just settled on controlling everything from a single app, and each device arrived as its own plug-in.
Left alone, every plug-in becomes a separate app inside the same app. A common screen frame and a shared set of visual elements are what stop that.
We made the plug-in graphics and visual identity together, and the scope grew into a further contract.
3 separate engagements on this line, 2016–2017.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In the mid-2010s, the connected home was a collection of brand apps. Each appliance maker shipped its own, and the obvious remedy was a single app that hosted a panel for each device. Samsung, which had agreed to acquire SmartThings in August 2014, went on to fold its separate home apps into the SmartThings app in 2018, and rivals pursued the same one-app idea through their own platforms. The model worked, but it worked inside walled gardens: a device appeared in one ecosystem's app because its maker had written for that ecosystem.
The decade since has been about loosening those walls. Appliance and HVAC makers formed the Home Connectivity Alliance, whose members include Samsung, LG, Electrolux, Haier, Midea and Trane Technologies, with a stated aim of an interoperable connected home. Its focus on large appliances complemented efforts aimed at smaller devices such as lights, locks and sensors.
The bigger shift came with Matter. The Connectivity Standards Alliance describes it as an IP-based protocol under which devices from different brands work together natively, running over Wi-Fi and Thread with Bluetooth Low Energy used for setup. SmartThings added Matter support in October 2022, as did the other major platforms. Matter 1.4, released in November 2024, extended Multi-Admin so that a device can join several ecosystems with a single user consent. The same lamp or thermostat can now sit in more than one household app at once.
This changes what a device panel is. Under a shared standard, the basic controls for a given device type are defined by the specification rather than drawn afresh by each manufacturer, which brings consistency and, at times, a lowest common denominator. Distinctive features still depend on the maker's own app or a platform-specific extension, so the layered model has not disappeared; it has moved. Platforms are also adding automation, energy reporting and natural-language assistants on top, so the interface people meet is increasingly a summary composed by software rather than a direct view of each device.
For the people living with these homes, interoperability solves one problem and introduces another. Fewer devices are stranded in the wrong app, but the same device may now answer to several controllers with different names, room assignments and routines. Users need to know which app holds authority over what, why a device changed state, and how to recover when one ecosystem and another disagree. The interface work has shifted from drawing each device well to keeping a household's mental model of its home coherent across all the surfaces that can touch it.

Quaduron Home Appliance for Commercialisation UX Guidelines Development

One platform underneath, and a different development organisation for every product on top of it.
Several appliance lines were being commercialised on a shared chip and platform.
A guideline written for organisations that do not talk to each other has to be exact about very little and silent about a great deal. What every product keeps, and what each is free to decide for itself.
We wrote the guidelines from the commercialisation point of view, and the scope — and the contract — grew during the work.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In the late 2010s, connectivity in household appliances was becoming standard equipment rather than a premium extra, typically supplied through common Wi-Fi modules and cloud services reused across many models. The economics were obvious. The less obvious consequence was that a weakness in a common component could become a weakness in every product that used it. Since then, the security of connected appliances has turned from a matter of good practice into a regulated property of the product.
The first broad benchmark came from European standards work. In June 2020 ETSI published EN 303 645, a baseline of 13 security provisions and 5 data protection provisions for consumer connected products, whose stated scope includes connected washing machines and fridges. It addressed the ordinary failures behind large-scale attacks on smart devices, such as shared default passwords, absent update routes and unclear handling of reported vulnerabilities, and it became a reference point for the legislation that followed.
Law followed within a few years. The United Kingdom's product security regime came into effect on 29 April 2024, banning universal and easily guessable default passwords and requiring manufacturers to publish how to report security issues and the minimum period for which a product will receive security updates. The European Union's Cyber Resilience Act entered into force in December 2024, with reporting obligations from September 2026 and its main requirements, tied to CE marking, applying from December 2027. In the United States the approach is voluntary labelling: the FCC named a new lead administrator for its Cyber Trust Mark programme in April 2026 and is still finalising how the label will work.
For manufacturers building many products on common components, these rules change the arithmetic. A support period is now a disclosed attribute of the product, sometimes required before sale. Update capability can no longer be an option fitted to premium models. It has to be designed into the common base from the start. Vulnerability handling becomes a lifecycle duty measured in years, which favours platforms designed from the start to be maintained rather than simply shipped.
Security now surfaces in the interface, and that is where people meet it. First-time setup that requires a unique credential, a clear statement of how long a machine will be supported, update prompts that do not interrupt a cycle, and a plain account of what data leaves the home are all user-experience problems as much as engineering ones. The design task is to make these things legible and routine, so that people understand them without being alarmed by them, and so that an appliance meant to last a decade does not quietly become the weakest device in the house.

CELL(WiFi router)Web app UX Design

Every device in the house can be measured — its size, its weight, its model number. The one thing nobody can see is what joins them.
A router is the device everything else depends on and no one ever looks at. It sits on a shelf doing its work invisibly, and the only time a person opens its settings is when something has stopped working — which is the worst possible moment to be met by a vocabulary.
SSID, band, channel, DHCP, WPA: these are not words anyone learns, they are words people are made to accept. So the work is how many steps can be taken out, and how much of that language can be said in plain words instead.
What the person wants to know is always smaller than what the interface offers — which things in the house are connected, whether each of them is all right, and what to do about the one that is not.
UX design for the router's web app, carried through an additional phase.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s the home router has gone through a quiet change of identity. In 2017 most households still had a single box supplied by their internet provider, configured, if ever, through a web page at a local address. Today the typical arrangement is more likely to be several units forming a mesh, managed through a phone app and updated in the background without anyone noticing.
Mesh systems made that shift visible. When Samsung launched SmartThings Wifi in 2018, it paired Plume's adaptive mesh technology with a built-in smart home hub, part of a wave of products that treated coverage and connected-device management as a single problem. Choices about channels and bands that once lived on an advanced settings page increasingly moved into software that adjusts itself.
Security standards followed. The Wi-Fi Alliance introduced WPA3 in June 2018, with a handshake designed to resist password guessing even when people pick weak passwords, and announced alongside it Wi-Fi Easy Connect, which lets a phone enrol a device without a screen by scanning a QR code. Regulators have since turned good practice into obligation. The UK's connected product security rules, in force since April 2024, bar easily guessable default passwords on consumer smart devices and require manufacturers to say how long security updates will be provided.
The radio layer, meanwhile, has become more capable and more complicated. Wi-Fi CERTIFIED 7, launched in January 2024, brought 320 MHz channels in countries that have opened the 6 GHz band, along with Multi-Link Operation, which lets a device use several links at the same time. Each generation adds options that almost no householder will want to tune by hand, while the number of things on a home network, from speakers and cameras to appliances and sensors, keeps rising.
The result is a device that is simultaneously more autonomous and more consequential. A household network now carries work calls, school, security cameras and door locks, and a fault in it can stop all of them at once, yet the equipment is designed to be forgotten.
For users, the direction is less configuration and more supervision. People are rarely asked to make technical decisions about their networks now, but they remain responsible for what is connected and whether it is safe. Interfaces serve them well when they make security defaults visible rather than silent, show when the system has changed something on its own, and offer a clear way back when automation gets it wrong.

UNICEF Web Audit

Someone who leaves halfway through a donation does not come back later.
The audit's first decision is where to look — the screens between intention and completion are worth more attention than the rest of the site put together. We diagnosed the problems and proposed the direction.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, online giving has followed shopping onto the phone, but the gifts have not quite followed the visits. The 2026 edition of M+R Benchmarks, which tracks US nonprofits, found that phones and tablets account for 52% of visits to charity websites, while desktop users still make 57% of donation transactions and a larger share again of revenue. People discover a cause on a small screen in a spare moment; many of them complete the gift somewhere else, or not at all.
Much of the work of closing that gap has moved from individual sites into the platform. The W3C's Payment Request API, which first became a Recommendation in 2022 and is still being refined, lets a page ask the browser to collect a payment method and, where needed, a name, email, phone number or address, with the browser acting as intermediary between payer, site and payment provider. Combined with wallets on phones, it can turn what used to be a dozen typed fields into a confirmation step, and it moves attention from form layout to the moments just before and after payment.
Accessibility standards have moved in the same direction. WCAG 2.2 added criteria that read almost as if they were written with donation flows in mind. Redundant Entry says a multi-step process should not make people supply the same information twice, such as a billing address already given for delivery; the W3C explains that recalling and retyping details wears down people with memory difficulties, and that leaning on browser autofill does not satisfy the criterion. Accessible Authentication (Minimum) rules out sign-in steps that depend only on remembering or transcribing something, and treats blocked pasting or disabled password managers as failures.
These rules matter more as charities encourage regular giving and donor accounts. A monthly gift is not a single form but a relationship: a login, a payment card that expires, a receipt, a way to pause or stop. Each of those touchpoints is another place where an inaccessible step or a forced re-entry can end the relationship quietly. Organisations that once looked at a single donation page now have to consider the whole life of a supporter across devices.
For donors, the trend points towards giving that asks for less memory and less typing at the moment of commitment, and that lets a decision started on a phone be finished on a laptop without starting again. For people with cognitive, visual or motor impairments, these details are not polish; they decide whether a willing supporter can complete the act at all. Good donation interfaces now treat payment as the shortest part of the journey and spend their care on clarity: what the gift does, what will be charged and when, and how easy it is to change one's mind.

Mappy OBN 3.0 UI Design

The map arrives over the air. Sometimes it does not arrive.
A connected navigation system fetches its maps and routes as it goes, so the screen spends part of its life waiting.
Waiting is a state to be designed, not a gap between states. What is shown while the data is on its way, and what is said when the connection goes.
We started from the embedded navigation work, listed the states that only exist in a connected system, and designed from that list.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In-car maps began as a storage problem. When the Navigation Data Standard association was founded in 2009, its aim was a shared format for in-car map data that was small and efficient, because maps had to fit on CD-ROMs, then DVDs, then hard drives. By 2020 the same association was citing Europe's eCall mandate, the start of 5G and the need for fresher data as reasons to rebuild the standard as a distributed system called NDS.Live, while keeping it backward compatible.
Since the late 2010s, that redesign has tracked where navigation has been heading. The association describes NDS.Live as supporting the shift from offline to hybrid and online navigation: map data delivered as tiles, along the driving path or as single features; updates limited to what has changed; components spread across the car, the network edge and the cloud; and thin data layers for EV charging, parking and zoning. Its members include BMW, Mercedes-Benz, Renault, Stellantis and Volvo alongside map suppliers such as TomTom and HERE. A car's map is no longer one database replaced now and then but a stack of layers refreshed at different speeds.
Some of those layers change daily. HERE, for example, offers a connected Speed Limit Layer that is updated every day as an add-on to its navigation map data, checking map values against detections from vehicle sensors and anonymised probe data collected under GDPR. Fresh data now comes partly from the fleet itself, which brings privacy and consent into what used to be a purely cartographic product.
Connectivity has not removed the offline case; it has turned it into a mode. Phone navigation shows the pattern. Google Maps' offline maps can guide a driver when the connection is slow or absent, provided the whole route lies inside a downloaded area, but traffic information and alternative routes are unavailable, and downloaded areas expire unless they are refreshed, usually over Wi-Fi. Built-in systems face the same balance between what is stored in the car and what is fetched on demand, and electric-vehicle routing leans on live layers such as charging data more than any petrol-era route ever did.
For drivers, the practical effect is that a map now has an age as well as a coverage. The same screen can combine data that is minutes old with data that is months old, and it is rarely obvious which is which. Interfaces that make the source and currency of information legible, and that signal plainly when guidance has fallen back to stored data, let people judge how far to rely on what they see. The more of the map lives in the network, the more that candour matters.

Smart Helmet Premium Value-Added Service Consulting

A rider cannot operate anything while riding. Almost the whole service has to happen before the ride and after it.
The helmet carries a camera, a speaker, a GPS receiver and sensors for position, altitude, speed and attitude. What comes off it is one ride recorded twice — as footage, and as numbers. The numbers are what make the footage usable: they know where the climb was, where the speed rose, where the machine leaned, and those are the places worth cutting to.
Everything the rider actually does with it, they do with their hands free — on a phone or on the web, after the helmet comes off. Editing the video, planning the next route, keeping the record of where they have been. The helmet's part is to capture all of it without ever asking to be operated.
And riding of this kind is social. The recording exists because it is going to be shown, so somewhere to show it is not an extra feature; a community site belonged inside the service rather than beside it.
We proposed the direction for the premium services and carried it through to the interface.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the connected motorcycle helmet has travelled a winding road. The emblematic early promise was a heads-up display built into the shell, and it ran ahead of what a young company could manufacture: Skully, whose AR-1 had been taking pre-orders, closed its doors in 2016 and filed for liquidation. The industry drew a cautious lesson from it. For several years the durable products were add-ons rather than all-in-one helmets: Bluetooth intercoms clipped to the shell, action cameras on chin mounts, and phones in handlebar cradles doing the computing.
Established helmet makers approached the idea more slowly. Shoei showed its IT-HT at CES in January 2019, projecting speed, engine revs and turn directions onto a small lens in front of the rider's right eye, and planned to sell it only in Japan. In late 2025 the company went further with the GT-Air 3 Smart, developed with the smart-technology firm EyeLights, which places navigation and safety information in the rider's line of sight. The six years between the two say a good deal about how hard it is to fit optics, a battery and wiring inside a shell that must still pass impact tests.
Regulation sits at the centre of that difficulty. When the UN Economic Commission for Europe approved the 22.06 revision of its helmet regulation in June 2020, the text required helmets to be tested with and without the accessories intended for them, and limited approval to the accessories actually tested. A rule like that favours electronics designed together with the shell over electronics bolted on afterwards, and it helps explain why the integrated models now reaching the market carry the names of long-standing helmet brands rather than start-ups.
At the same time, some of the safety intelligence once promised for helmets has settled in devices riders already carry. Apple's 2022 iPhones, for instance, can detect a severe car crash and dial emergency services when the owner cannot, using motion sensors and algorithms rather than dedicated hardware. Automatic crash alerting, live location sharing and rider-to-rider intercom groups now turn up across helmets, communicators and phones alike, and the boundaries between those products keep shifting.
For riders, the open questions are about attention and trust. Anything shown inside a visor competes with the road, so the useful designs keep in-ride information sparse, steady in position and quick to read, and let audio carry what does not need to be seen. Automated safety features raise a different demand: a system that may call for help on its own has to make plain what it has detected, give the rider a clear and simple way to cancel a false alarm, and behave the same way every time, because a rider will only rely on what they can predict.

Company-wide App Icon/Branding Element Design Development

The same app wore a different icon and a different colour on every device.
A brand is not applied at the end; it is the thing that has to be identical in four places at once. We analysed the icons, colours and layouts of the apps common to phone, watch, TV and Family Hub, wrote the standard, and made the film that introduced it internally.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the app icon has stopped being a finished picture. In 2017 an icon was, for most purposes, artwork that a designer exported at a range of sizes, and consistency across products depended on guidelines and discipline. Over the following decade the platforms took more of the rendering into their own hands, and the icon became a set of layers and rules that the system assembles at the moment of display.
Android moved first. Android 8.0 introduced adaptive launcher icons, which support system visual effects and can take different shapes on different device models. The app supplies foreground and background layers, and the launcher decides the outline. Android 13 went further with themed app icons: an app that provides a monochrome version lets the system tint it to match the user's wallpaper and theme, falling back to the ordinary icon where support is missing.
Apple's change came later but went deeper. In June 2025 it introduced Liquid Glass across iOS 26, macOS Tahoe 26 and its other platforms, with icons that render in light, dark, tinted and clear appearances and a developer tool, Icon Composer, for producing them. The first year was not smooth. Many users found the icons blurry and over-glossed, and by the iOS 27 developer beta in June 2026 Apple had rebuilt the glass as distinct layers within the artwork and calmed the tilt-driven shimmer.
The common thread is a transfer of control. Brand owners still draw the mark, but the platform now decides its outline, its colour in a given theme, its material and how it responds to light. The same icon may appear in a user's chosen tint on one device, in a glass finish on another and as a monochrome glyph on a watch. Consistency is no longer achieved by making every file identical. It is achieved by making the source material robust enough to survive transformations the brand does not control.
That raises the bar on the drawing itself. A mark that relies on one particular colour may lose its identity when tinted; one that depends on fine detail may disappear under a glass treatment or at watch scale. Teams increasingly test icons across themes, shapes and sizes in the way they test layouts across screens.
For users, the risk is recognition. Nielsen Norman Group has long found that only a handful of icons are understood without a label, and that meaning erodes when the same symbol behaves differently in different places. As systems recolour and reshape icons to suit a theme, the cues people use to find an app, colour above all, become less dependable. Good practice follows from that: keep silhouettes distinctive, keep labels visible, and treat consistency of meaning as more important than consistency of finish.

Byxbyhome SDK portal UX

A guideline site for the developers writing the software inside Samsung's appliances.
Two rulebooks had to become one — Android's reference guidance and Samsung's own One UI policy — and the portal is where a developer finds out which of them applies to the thing in front of them. The scope grew twice during the work.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, design guidance for software platforms has been changing from something people read into something tools consume. In 2017 a guideline portal was largely a collection of pages, diagrams and downloadable kits; designers and developers read it and translated it into their own work by hand. That model still exists, but the centre of gravity has moved towards structured, versioned data that design tools, build pipelines and now AI coding agents can read directly.
The scope of the guidance has widened at the same time. Samsung's One UI design guide now presents one system spanning phones, tablets, watches, earbuds and PCs, organised around four principles and backed by sections on layout, components, colour, iconography, motion, writing, accessibility, and large-screen and foldable devices. Google has kept evolving Material Design as well; in May 2025 details of Material 3 Expressive surfaced ahead of Google I/O, an iteration on Material You with bolder use of shape and colour that Google said was grounded in user research.
The more fundamental change is in the form guidance takes. In October 2025 the W3C Design Tokens Community Group released the first stable version of its Design Tokens specification, a vendor-neutral format for colour, typography and other design decisions. It supports light and dark modes, accessibility variants and multiple brands without duplicated files, and a single token file can generate platform code for iOS, Android, the web and Flutter. It is a community report rather than a formal W3C standard, but it gives tool makers a common format, so that a decision changed once can reach every platform that depends on it.
AI coding tools have added urgency. Figma's MCP server, introduced in beta in 2025, passes components, styles and variables from design files to coding agents and can produce a rules file describing a team's tokens and naming conventions, on the premise that generated code is only as consistent as the design-system context it is given. When a growing share of interface code is drafted by software, guidance that exists only as prose is guidance that will be missed.
None of this removes the need for a portal people can read. Developers still need intent, examples and the reasoning behind a rule, and they need to know when guidance has changed. What shifts is the portal's role: less a library of pages, more the explanatory front end to a body of machine-readable decisions, with version history and change notes treated as first-class content.
For the developers who rely on such guidance, the qualities that matter most are findability and currency: the right component for the right device, in its latest form, with the reasoning attached. For the people who eventually use the resulting products, the benefit is indirect but real. Interfaces behave consistently across a household of devices because the design decisions reached the code intact.

SLEEPsense 2.0 UI

The sensor sits under the mattress and is never touched. The report in the morning is the entire product.
A contactless sensor means the user has no relationship with the device itself — only with what it tells them.
A night produces more data than a morning has patience for. What leads at breakfast, and how much of it belongs in the report a family member sees.
We went through the 1.0 flow first and rebuilt the interface around the structure of the report. On a shared engagement, the interface design was our part.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, sleep tracking has developed along two lines: wearables that people put on at night, and contactless sensors that sit under the mattress or on the bedside table. The second line promised data without effort, with nothing to charge on the wrist and nothing to forget. Samsung was an early entrant. It unveiled SleepSense at IFA 2015, an under-mattress sensor developed with the medical device maker EarlySense, which produced a sleep summary in the morning and offered a family-care option for keeping an eye on an older relative's sleep.
Radar has since joined pressure and motion sensing. Google's second-generation Nest Hub, launched in 2021, used a low-power Soli radar chip to follow movement and breathing from the nightstand, processing the motion data on the device and validating its algorithms against polysomnography. The appeal was the same as before, no wearable, but with an explicit privacy argument attached, since a radar sensor produces movement traces rather than images.
The most significant shift of recent years is regulatory. Consumer sleep products long stayed on the wellness side of the line, offering scores rather than findings. That boundary has started to move. In February 2024 Samsung announced that a sleep apnoea feature for the Galaxy Watch had received De Novo authorisation from the US FDA, following approval by Korea's Ministry of Food and Drug Safety in October 2023. It looks for signs of moderate to severe obstructive sleep apnoea over two nights in adults who have not been diagnosed. Apple received US clearance for a comparable notification feature later the same year. Contactless sensing crossed the same line: in September 2024 Withings announced FDA 510(k) clearance for its Sleep Rx under-mattress mat as an aid to diagnosing obstructive sleep apnoea at home, offered in the US through clinical and remote-monitoring programmes.
The fine print matters. Samsung states that its feature does not replace diagnosis and treatment by a clinician and is not intended for people already diagnosed, and the Withings mat is cleared as an aid rather than as a diagnosis in itself. The direction is towards consumer hardware that can raise a flag worth taking to a doctor, not towards devices that replace a sleep study. The result is a widening gap between a wellness layer of scores, stages and routines and a narrow, regulated layer of screening results that carry clinical weight.
For the people reading these reports, the difficulty is telling the two layers apart. A wellness score and a screening notification can appear in the same app while meaning very different things, and a contactless device offers no physical interaction through which a user might learn its limits. Clear separation between advisory and clinical information, plain language about uncertainty, and an obvious route to professional advice are becoming as important to these products as the sensing itself.

Samsung Internet PC Extension UX

Bookmarks made on a phone, wanted on a desktop, inside somebody else's browser.
An extension lives in a window it does not own, so what it may do and when it may appear are decided by the browser before they are decided by us. UX for the PC extension.

App Analysis and Standard GUI Layout Guide

A home screen that arranges itself around the hour, the weather, where you are and which country you are in.
Concept work for a common mobile home screen, with the guidelines for the widget screens beneath it.
A template that adapts has to decide what it is allowed to assume. Guess well and it reads as attentive; guess badly and it reads as guessing, which is worse than a screen that simply sits still.
Built on an analysis of the existing apps and the layouts they had each settled on.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the late 2010s, the home screen has gradually stopped being a grid of app icons and become a surface for information. In 2017 widgets were largely an Android habit, and most people still reached what they needed by opening an app. Over the following years the major platforms converged on a different idea: the first screen a person sees should already show something useful.
Apple's turn came with iOS 14 in 2020, which let widgets sit on the Home Screen in several sizes and introduced the Smart Stack, a pile of widgets that uses on-device intelligence to bring forward the right one according to time, location and activity. Android answered with a broad refresh in Android 12: widgets gained responsive layouts, device-theme colours and interactive controls such as checkboxes and switches, so a widget could complete small jobs instead of only displaying a summary.
The next step moved information further forward, onto the lock screen, and handed more of the selection to AI. With the Galaxy S25 in January 2025, Samsung introduced Now Brief, which the company says learns a user's routines and delivers tailored information to the lock screen, supported by an on-device Personal Data Engine. The emphasis on on-device processing is telling. A surface that reflects someone's routines has to know those routines, and vendors now present privacy as part of the feature rather than a footnote.
The other question is control. Nielsen Norman Group draws a useful line between customisation, which the user configures, and personalisation, which the system decides. The first costs effort that many people never spend; the second depends on how well the system infers what someone needs, and can feel intrusive when it seems to know too much. Contextual home screens sit squarely on that line, and the better ones now offer both: an automatic stack or briefing, with settings to pin, remove or reorder what appears.
Design practice has followed. Widget guidance now specifies ranges of sizes rather than fixed cells, behaviour in light and dark themes, and how a widget looks before and after the user configures it. A home screen layout has become a system of slots whose contents change, and designing it means writing rules as much as drawing screens.
For users, a screen that changes with the day offers less searching at the price of less predictability. The interfaces that serve people best tend to keep a stable frame while the contents rotate, make it clear why an item is being shown, and give a quick way to dismiss or keep it, so that the screen adapts to its owner rather than training its owner to adapt to it.

Printing App Center 2.0 UI/GUI Development

A printer on every floor, a different one in every department, and someone who just wants a document.
The machines are not the problem; the difference between them is. Interface and graphics for version 2.0 of the app centre that holds a whole building's printing together.

Improvements to the CSS HMI on Amphibious Assault Ship (LPH)

ROKS Marado (LPH-6112) is the second Dokdo-class amphibious assault ship of the Republic of Korea Navy.
ROKS Marado is the second and final ship of the Dokdo class, built by Hanjin Heavy Industries in Busan and commissioned in June 2021. Like its sister ship ROKS Dokdo, it carries marines, vehicles, landing craft and helicopters, and its command support system supports navy, army, marine and allied units taking part in an amphibious operation.
AEON carried out HMI improvements for the ship's command support system.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
An amphibious assault ship is as much a floating headquarters as a transport. Ships such as South Korea's Dokdo class embark staff from naval, ground and air forces and run the command and control that ties helicopters, landing craft and troops ashore into one operation. ROKS Marado, the second of the class, was commissioned in June 2021 with an upgraded sensor fit and a flight deck able to take the V-22 Osprey.
Multi-service command at sea is increasingly a networking problem. The US Navy's Project Overmatch, described as the service's contribution to the Pentagon's combined joint all-domain command and control effort, has been rolled out to carrier strike groups, amphibious ready groups and maritime operations centres. Its purpose is to let dispersed ships and staffs share data across services, so that an amphibious group is one node in a wider joint network rather than a self-contained command post.
Ashore, joint headquarters are becoming permanent. Japan established a standing Joint Operations Command in March 2025 to direct its ground, maritime and air forces across all domains in real time and to work alongside US forces, replacing earlier ad hoc arrangements. Permanent joint commands raise the expectation that a ship's command support system can exchange plans and a common picture with national joint headquarters and allied staffs, not only with the units it carries.
The largest change is in what flies from the deck. Turkey's TCG Anadolu, a landing helicopter dock, launched and recovered a Bayraktar TB3 drone in November 2024, and TB3s are intended to be a central part of its air wing, providing reconnaissance, strike and communications relay. South Korea is moving in the same direction: its navy plans to upgrade Dokdo and Marado for drone operations and manned-unmanned teaming command, and to build a larger drone-capable ship for service in the late 2030s. Amphibious ships are becoming hubs for uncrewed aircraft and surface vessels as well as helicopters and landing craft.
For the people aboard, this widens the job considerably. A command staff drawn from several services, and sometimes several nations, must now coordinate drone sorties, joint fires and network-shared information as well as the traditional ship-to-shore movement. Each service brings its own vocabulary and priorities, so command support systems need shared pictures that every staff can read, views suited to each role, and clear handover of control between ship and shore as an operation moves inland.

Smart Home Mobile App UX Prototype

Checking the experience before committing to build it.
A prototype is defined by what it is allowed not to do, so scoping what would actually be tested came before making anything. We built the core scenarios far enough to run, and the direction could be judged on them.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In 2016 the smartphone was the smart home's remote control. Most connected devices arrived with an app of their own, and platform apps such as SmartThings and Apple's Home, launched that autumn, tried to gather them into a single grid of tiles. Voice was only just entering the home: Amazon's Echo was spreading, and Google Home went on sale at the end of the year.
The early voice assistants promised more than they delivered. A 2018 Nielsen Norman Group study of Alexa, Google Assistant and Siri found their usability fell well short of what was promised: they handled simple, short requests but stumbled on complex ones, cut people off when they paused mid-sentence, and left users to work out exactly how a command had to be phrased. For the smart home, that meant voice worked best for brief, well-rehearsed commands, while anything more elaborate, such as schedules, scenes or conditional routines, was still built on a phone.
Generative AI has shifted that balance. Amazon's Alexa+, announced in February 2025, holds open-ended conversations, lets people create routines entirely by voice and can carry out multistep tasks on their behalf. Google's Gemini for Home, announced in August 2025, is gradually replacing Google Assistant on its speakers and displays and accepts compound requests with exceptions, such as switching off every light except those in one room. Configuration that once demanded an app is moving into conversation.
The interoperability work of the same period supports the change, since shared device standards give assistants a common description of what they are acting on. The phone app has not disappeared, but its role is changing: it is becoming less the place where every action happens and more the place where people set up, review and correct what the home does on its own or on a spoken instruction.
For residents, the move from tapping tiles to delegating tasks changes what good interaction looks like. When a system acts on a loosely worded request, people need to see what it understood, what it did and how to undo it; they need confidence about what a microphone or camera keeps; and households need ways to share control without one person's instruction silently overriding another's. Directness used to be the measure of a smart-home interface. Increasingly, the measure is legibility.

Smart Building Service UI/GUI Development

Three shifts a day, and not one of them spends its time watching the screen.
In a large building the manager has to see communications, security and climate in one place. Nobody can keep looking at it.
So the information divides in two: what must be legible whenever someone glances up, and what has to be caught in the instant it happens. They are different design problems and they compete for the same screen.
We gathered what the building's people needed and found difficult, then designed an interface that put visibility first because monitoring is what it is for. This engagement led on to the commercialisation of b.IoT.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
In the mid-2010s, integrating a large building's subsystems into a single operations platform was mostly argued as a matter of efficiency and staffing. A decade on, the same convergence is just as often discussed as an attack surface. Building systems that were engineered for isolated networks now share infrastructure with corporate IT, remote maintenance links and cloud services, and the boundary between facilities and cyber security has largely dissolved.
The case most often cited is a retail breach from 2013. According to reporting at the time, attackers entered the retailer's network with credentials stolen from a refrigeration and HVAC contractor, and the episode raised the question of why a contractor's access was not separated from payment systems. The building equipment itself was not the target, which is precisely why the story stuck: it showed that the plumbing of building operations could serve as a route into everything else.
Protocols have been catching up. BACnet, the dominant building automation protocol, gained a secure option known as BACnet/SC, which uses TLS with certificate-based mutual authentication and does away with broadcast traffic, so that devices can protect themselves on shared IP networks without separate VPN equipment. Adoption is gradual, since controllers stay in service for many years, but new specifications increasingly ask for it.
Vulnerabilities in building controllers continue to surface, and the official advice is consistent. A 2025 US advisory for one family of building controllers repeats the standard guidance to keep control devices off the internet and place them behind firewalls, isolated from business networks. Regulation is adding obligations on the supply side: the EU Cyber Resilience Act, which covers networked hardware and software, entered into force in December 2024, with manufacturer reporting duties in effect since 11 September 2026 and full requirements due in December 2027. Building operators are also absorbing practices from IT, including identity management for contractors, logging of configuration changes and, in larger organisations, operations centres that watch facilities and network security side by side.
For the people on shift in a building operations room, this means security events now sit alongside comfort complaints and equipment faults in the same working day. The interface has to show who changed what, from where and with what authority, and it has to make an unusual remote session as noticeable as an unusual temperature. At the same time, authentication cannot become so cumbersome that staff share logins to get through a night. Designing for accountable, individual access that still fits round-the-clock operations is now part of the usability brief, not an add-on to it.

Smart Home Tablet UX Prototype

A tablet can show the whole house at once, which is not automatically an advantage.
Several devices on one large screen is a question of arrangement, not capacity. About a month to build the tablet UX prototype.

Samsung Members of UI and Android UI GUIDE

Diagnostics, customer support and a community forum, in one app.
Three functions with nothing in common need a structure that does not pretend they have something in common. Interface design for the app, with the Android UI guide alongside it.

Convergence App UI/GUI Design

An app that moves between devices has to explain the move without explaining it.
A connection is understood or it is not; there is no halfway that a caption can rescue. Interface and graphics, developed jointly, with the scope growing during the work.

Message PC App UI/GUI

The same conversation, continued on a bigger screen.
Messages sent from a phone, read and answered on a PC. What has to stay identical is the thread itself — a conversation that looks like a different conversation on the other device is not continuous, whatever the data says. Interface and graphics developed together.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, sending a text from a computer has gone from a vendor-specific convenience to an expectation on every major platform. In 2016 the experience depended on which phone, operating system and carrier a person happened to use: iMessage already appeared across Apple's own devices, while on Android it varied from one manufacturer to the next. A decade later, the plumbing underneath has changed more than the screens on top of it.
Google moved first on the Android side, launching Messages for web in June 2018: a browser tab paired with the phone by scanning a QR code, relaying texts, stickers and images through the handset. Microsoft took a similar route with Your Phone, later renamed Phone Link, and in 2023 extended it to iPhones, though with clear gaps: no replies to group chats, no photos or video, and messages that only arrive while the phone is connected. Those limits illustrate a recurring pattern. A desktop client that relays the phone inherits every constraint of the link between them.
The larger story has been the protocol. RCS, the carrier-backed successor to SMS, spent years with patchy support while Google pushed it through its own Messages app. In 2024 Apple added RCS to iPhone Messages with iOS 18, bringing richer media and more dependable group chats to conversations with people who do not use Apple devices. For the first time in years, the two largest phone platforms shared a modern messaging baseline.
Security followed. The GSMA's RCS Universal Profile 3.0, published in 2025, defines end-to-end encryption built on the Messaging Layer Security protocol, and in May 2026 Apple and Google began rolling out encrypted RCS between iPhone and Android in beta, with a lock icon marking protected conversations and support depending on carriers. For companion apps on computers, that raises the bar: a conversation encrypted end to end has to remain so on every device allowed to display it, and pairing a new screen becomes a matter of trust as well as convenience.
For people moving between a phone and a computer, the significant change is that messaging is converging on shared standards rather than on any single vendor's app. That shifts the interaction questions toward transparency: showing which protocol a conversation is using, whether it is encrypted, which devices can read it, and which features will not be available on the screen in front of them. Most users never think about protocols, but they notice at once when something that worked on one device is missing on another.

Energy Reduction Solution Promotion for Demo Application

A few minutes at a stand, in front of someone who may not be technical.
The energy solution had to show what it saves, quickly, at sales meetings and exhibitions.
A demonstration is an argument about which scene to show. Pick the function that saves the most and it may take too long to read; pick the one that reads instantly and it may not be the point.
We chose the moments where the saving is visible on its own and built the demo application around them.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, the way energy savings are presented has shifted from projection towards proof. A saving used to be something a vendor demonstrated and a buyer took largely on trust. Today the same figure is more likely to be audited, reported to a regulator, written into a performance contract or checked by a tenant with access to the meter data. The audience for an energy number has widened, and so has the scrutiny.
Part of the difficulty is that a saving is not a quantity anyone can read off an instrument. The Efficiency Valuation Organization, which maintains the main international protocol for measurement and verification, points out that savings cannot be measured directly, because they are the absence of consumption, and must be found by comparing measured use before and after with adjustments for changed conditions. Weather, occupancy and operating hours all move the baseline. Credible reporting has therefore come to mean normalised figures and stated assumptions rather than a single before-and-after bar.
Research on feedback has also tempered expectations about what a display alone can achieve. A meta-analysis of 42 studies in Psychological Bulletin found that energy feedback works overall but with a small and widely varying effect, depending on frequency, medium, comparison and duration. Showing people their consumption helps most when it is timely, set against a meaningful reference and connected to something they can actually change.
Policy has made verified performance a requirement in more places. New York's Local Law 97 applies emissions limits to most buildings over 25,000 square feet; the first reports, covering 2024, were due at the end of 2025, and the city says 93 per cent of covered properties filed. In Europe, the Smart Readiness Indicator rates a building on how well it optimises energy use, adapts to occupants and responds to signals from the grid. It remains voluntary for now, but it frames savings as a capability of the building rather than a one-off result. Flexibility, meaning when a building uses energy as well as how much, is becoming part of the story alongside efficiency.
For the people who read these figures, including facility managers, finance teams, tenants and auditors, the interface carries more responsibility than it used to. A saving shown without its baseline invites doubt; a saving shown with too many caveats loses its audience. The emerging practice is to pair a clear headline with an inspectable path behind it, so the same view can persuade a newcomer and satisfy a sceptic, and to link each figure to the actions that produced it.

North America Luxury Built-in UI Guidelines

A built-in appliance is furniture first. In this market it is also a claim about the kitchen it sits in.
North American luxury buyers expect a great deal, and the product has to belong to the room around it.
A guideline can specify a tone, which is harder than specifying a layout. What makes a screen read as premium, and how that is written down so that different products arrive at it independently.
We set out the market conditions and wrote them up as UI guidelines for the built-in range.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The North American luxury built-in market was long defined by heritage brands with decades of reputation in professional-style cooking. In 2016 two Korean manufacturers moved into it at once. Samsung agreed to acquire Dacor, a California maker of luxury kitchen appliances since 1965, and LG introduced its ultra-premium built-in line, Signature Kitchen Suite, in July of the same year. Both brought large consumer-electronics software organisations into a segment that had been shaped by cooking performance and craftsmanship.
Since the late 2010s, the clearest direction in this segment has been integration and concealment. Panel-ready columns, handle-free fronts and dishwashers with controls hidden on the top edge of the door all let the appliances recede into continuous cabinetry. Samsung's Dacor brand took this to its logical end at KBIS 2026 with a “hidden kitchen” concept placing a refrigerator, cooktop and dishwasher inside curved columns or furniture, while keeping SmartThings connectivity. The interface follows the cabinetry: displays that are invisible until touched, controls that reveal themselves only when needed, and status that appears briefly and then withdraws.
Cooking itself has changed surface. Induction has gained ground in North American kitchens, and the US Department of Energy notes that it heats the cookware directly through an electromagnetic field and can be up to three times as efficient as gas. For interface design this matters more than it first appears. An induction hob is a flat sheet of glass with touch controls, so the main cooking control in a premium kitchen is now a printed or illuminated surface rather than a knob. There is no flame to watch, and heat level, residual warmth and pan detection all have to be shown by light and symbol.
Connectivity has arrived in this tier too, and it creates its own tension. A built-in range is chosen to stay in place for many years, with a long design life, while the phone app and cloud service it depends on may change many times in that period. Luxury makers have to reconcile the slow, craft-led expectations of the room with the much faster update cycles of software. A remote alert that the oven has reached temperature adds convenience; a basic function that depends on an account or a server adds a point of failure that the cabinetry around it does not have.
For the people who cook in these kitchens, concealment brings its own usability problems. When controls hide, discoverability is the first thing at risk, so a hidden interface has to be easy to wake and obvious once woken. When there is no flame or visible element, feedback has to carry more of the burden. Safety states, such as a hot zone after the cooktop is switched off, must stay visible even when everything else is hidden. And because these kitchens are kept for a long time, controls at the appliance itself need to remain complete and understandable on their own, long after the companion app has changed.

Family hub 2.0 GUI Guidelines (Mobile App)

Everyone in the house uses the refrigerator, in passing, without deciding to. In 2.0 the phone joined it.
Family Hub tied the fridge screen to a mobile app, so a task begun in the kitchen could be finished in another room, by someone else.
The same function shown twice has to be recognisably the same function — and some of it should not be shown to everyone in the house. Both were decisions for the guideline, not for the individual screens.
We wrote the mobile app GUI guidelines on the appliance experience built up since the internet refrigerator. The scope grew during the work and the contract with it.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
When Samsung introduced Family Hub at CES 2016, it placed a 21.5-inch screen on the refrigerator door for shared calendars, notes and photos, with cameras inside that photograph the contents each time the door closes. The idea was that the kitchen already worked as the household's noticeboard and the refrigerator could become its digital version. Since the late 2010s that idea has spread well beyond refrigerators: smart displays from Amazon and Google took up similar places on kitchen counters, and the communal screen became a recognised category of device.
A communal screen has an awkward property. It has many users and, in the interface sense, no single owner. Early communal screens tended to treat the household as a single user. The direction since then has been towards recognising who is standing in front of the device. In 2019 Samsung added Bixby voice ID to Family Hub so that the refrigerator could tell family members’ voices apart and give each person information suited to them. Across the industry, voice and camera recognition have increasingly been used to personalise screens that the whole household shares.
HCI research suggests that the hard part is social rather than technical. A survey by Kraemer and colleagues, published in the CHI 2020 extended abstracts, highlighted that expectations about communal technology in the home depend on the qualities of the relationships between the people sharing it as well as of the technology, and proposed that designs draw on information such as location and levels of trust. The circle also extends beyond the household: a survey by Park and colleagues found nearly half of smart home owners had shared at least one device with someone who did not live with them, often a trusted relative given remote access.
More capable conversational assistants on these devices add a further layer. Kitchen displays and refrigerator screens increasingly answer open questions, suggest recipes from what is in stock and read out schedules, and an assistant that speaks to a room speaks to everyone in it. The industry's answer so far has been recognition, by voice or by face, but recognition can fail, and households differ widely in how much they want a shared device to know about each member. Account models built for personal phones fit awkwardly onto a device that a family, a visiting grandparent and a cleaner may all touch in the same day.
For the people who share these devices, a few needs recur. They need to see who the device currently thinks it is serving, and to have sensible behaviour when it recognises nobody. They need recognition errors to be cheap, so that being mistaken for someone else reveals little and is easy to correct. And they need a device whose behaviour fits the household's own sense of who may see and change what, rather than a single owner's settings imposed on everyone else who lives there.

Premium Air Purifier (AX7700) GUI/VI Development

It stands in the living room, so its display is part of the furniture — and it has to be readable from the sofa.
A premium air purifier is looked at from across a room, by people who are not thinking about it.
Air quality has to be legible at a distance and unobtrusive at rest, which are opposite demands on the same display. Colour and marking carry both: what dust and gas readings look like, and what the product's visual identity is when nothing is happening.
Graphics and visual identity were developed together, and the scope grew twice during the work.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, the definition of clean air has tightened, and that has quietly changed what an air purifier's display ought to say. In 2021 the World Health Organization issued new air quality guidelines with lower recommended levels for key pollutants, citing evidence of harm at lower concentrations than previously understood, alongside interim targets for countries with high pollution. In the United States, the Environmental Protection Agency moved the boundary between “Good” and “Moderate” on its colour-coded Air Quality Index for fine particles from 12.0 to 9.0 micrograms per cubic metre, effective May 2024. A reading that a device would have shown as comfortably good in 2016 may now sit in a band that public agencies call moderate.
At the same time, sensing has become cheap and widespread. Optical particle sensors now appear in purifiers, standalone monitors and community networks, and their readings increasingly sit beside official data. Their accuracy has been studied closely. EPA researchers found that raw readings from a widely used low-cost sensor network overstated fine-particle levels by about 40 per cent across most of the US, and that a correction including humidity brought them much closer to regulatory monitors. The lesson for device makers is that a precise-looking number is not the same as an accurate one, and that humidity and calibration shape what a display should claim.
The pandemic changed the vocabulary of indoor air as well. In July 2023 ASHRAE published Standard 241, which sets a target of equivalent clean airflow per occupant, combining ventilation, filtration and air cleaning into a single measure. That moved the conversation from whether a room's air is clean at a given moment towards how much clean air is being delivered to the people in it, a framing close to the clean air delivery rate long used to compare portable purifiers.
Colour remains the main channel for all of this, and it is where consistency is weakest. Public indices use established colour bands, while consumer devices can use their own schemes with their own thresholds. A living-room purifier glowing one colour while a phone weather app shows another for the same air leaves people unsure which to believe. Colour alone also fails people with colour-vision deficiencies, so position, shape or text has to carry the same message.
For people living with these devices, the practical consequences are fairly clear. A scale should agree with the public indices people see elsewhere, or say plainly where and why it differs. A smoothed trend is usually more useful than a jittery instantaneous figure, and a device should be honest about the limits of its sensor rather than implying laboratory precision. And because guidance on what counts as clean air keeps moving, the thresholds behind a display's colours are worth being able to update after sale, not fixed for the life of the product.

Mediana Patient monitor (Hawk) UI/GUI Development

The ward already knew how to read a patient monitor. The risk in redesigning one is not that the new screen is worse — it is that it is different.
A patient monitor gets read by people who are not looking at it. A nurse crossing the ward takes a number in and keeps walking. A doctor glances at it mid-sentence. That reading is not learned from a manual — it accumulates over years, on whatever machine the ward happens to own, and a new design inherits all of it.
So the first decision was not what to change. It was what to leave alone. Which readings hold their position no matter what else moves. How an alarm reaches someone who is facing the other way. And where the existing way of working is genuinely inconvenient, as opposed to merely unfamiliar to whoever is redesigning it.
We kept the learned environment and changed only where the inconvenience was clear, one function at a time. Units, colours and symbols were set against international standards rather than preference. On a device like this a colour is not a matter of style — it is the thing someone acts on.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, the most consequential changes in patient monitoring have had less to do with new sensors than with how monitors interrupt people. Alarm fatigue, in which clinicians become desensitised to frequent and mostly non-actionable alarms, was by then a recognised safety problem. ECRI Institute put alarm hazards at the top of its annual list of health technology dangers in late 2013. The same year the Joint Commission in the United States approved a National Patient Safety Goal on clinical alarm safety, phased in from 2014, which required hospitals to make alarm safety an organisational priority, identify their most important alarm signals and, from 2016, put alarm-management policies and staff education in place.
That moved attention from the individual device to the alarm system as a whole: default limits, delays before an alarm sounds, which alarms escalate to a central station or a nurse's phone, and who is allowed to change the settings. Much of the work that followed concerned configuration and clinical practice as much as hardware.
The sounds themselves have changed too. IEC 60601-1-8, the international collateral standard for alarm systems in medical electrical equipment, received a second amendment in 2020. It drew on research led by Judy Edworthy at the University of Plymouth, which tested candidate sounds against the older melodic alarm set. According to the university's impact case study, that work showed auditory icons, alarm sounds that suggest their meaning instead of relying on abstract tones, performing better than almost all alternatives, and those icons were incorporated into the standard. Because hospitals replace equipment gradually, wards will carry both generations of sound for some years.
Connectivity is the other long-running thread. Monitors increasingly feed central stations, electronic records and mobile alerting systems, and the IEEE 11073 service-oriented device connectivity (SDC) family of standards aims at manufacturer-independent communication between point-of-care devices. Continuous monitoring has also been spreading from intensive care to general wards and, through wearable sensors, to patients at home. Each extension adds signals competing for the same limited attention, and the evidence on which combinations improve patient outcomes is still developing.
For clinicians, the trend moves the design problem from a single screen to the stream of alerts that reaches them across a shift, from bedside monitors, pumps, ventilators and handheld devices. Trust grows when an alert is rare enough to be believed and specific enough to act on, and when sounds, priorities and escalation paths behave consistently across equipment from different makers. Human factors work in this field is increasingly about the whole system of notifications rather than any one device.
PC Smart Manager UX Design
People open this app because something is wrong. They are not in the mood to learn vocabulary.
A diagnosis that has to be looked up has not been delivered. The work was saying what the check found without the terms it was found in. UX design, completed in a short period.

Standard Gen 5 Navigation

Another language, another address system, another set of road rules. A screen built for one country does not simply translate.
Localisation divides into two piles: the words, and the structure. Confusing them is how a specification either bloats or breaks — everything treated as structure means a second product, everything treated as words means a screen that lies about the roads.
Working from the common Standard Gen 5 design, we pulled out what the local conditions actually change and wrote those into the specification.
2 separate engagements on this line in 2016.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Since the mid-2010s, the navigation screen in a car has changed hands more than once. In 2016 most embedded systems still shipped with a regional map database installed at the factory and refreshed, if at all, during a dealer visit or from a USB stick. Phone projection through Apple CarPlay and Android Auto was spreading quickly, and for many drivers the phone rather than the car became the navigator they trusted.
Carmakers have since answered in opposite directions. In 2023 General Motors said it would phase out CarPlay and Android Auto in its electric vehicles in favour of built-in software based on Google's Android Automotive, keeping the dashboard under its own control. Others opened the door wider. When Apple began rolling out CarPlay Ultra in May 2025, a version that extends the phone's interface into the instrument cluster and to controls such as climate and radio, Hyundai, Kia and Genesis were among the makers Apple named as committed to it.
Embedded navigation has meanwhile caught up on freshness. Hyundai's ccNC cockpit now receives map updates over the air every month, including new roads, changed routing and local speed limits, downloading only the relevant sections in the background while the car is driven. The map has turned from a file shipped with the car into a service that keeps changing, and that alters what localisation means: local speed limits and routing can now follow a country's changes instead of freezing at the model year.
Maps also remain national infrastructure, and that shapes navigation as much as language does. South Korea is a sharp example. Having refused Google's requests to take detailed map data abroad in 2007 and 2016, the government approved a request in February 2026 only under conditions, including processing on servers in the country, government review before export, use limited to navigation and the removal of military and security sites. Any platform that wants to guide drivers across borders has to live within rules of this kind.
Testers have turned their attention to the driver's hands and eyes. Euro NCAP's 2026 protocols score how easily basic functions can be operated, and its first results in October 2026 favoured cars that keep indicators, wipers, hazard lights, the horn and the emergency call button on physical controls, while a model that relies on its touchscreen for basic functions scored poorly on that measure.
For drivers, the result is that the dashboard is no longer one product but a meeting point between the carmaker, the phone platform and the map supplier. That raises the bar for consistency: the same instruction should look and sound the same wherever it appears, and controls that matter in an emergency should stay where a hand can find them without a glance. Localisation now lives in the whole interaction rather than in a map file, and it has to survive a steady stream of updates from more than one source.
Mobile Print App 4.0
Nobody opens a printing app to look at it. They open it because a sheet has to come out.
Version 4.0 had to show the printer's state and its options so that both read at a glance, which is a different task from listing them.
So the only number that matters is how many steps stand between opening the app and the sheet coming out.
Interface and graphics developed together. 2 separate engagements on this line in 2015.
Galaxy Apps
Store interface for Galaxy Apps — how apps are browsed, and how browsing turns into installing.
Two stores, one errand. The design was the join — how looking for a Gear app on a phone leads to it being installed on the watch without the user tracking where anything is. Produced together, with the period extended.
3 separate engagements on this line, 2014–2015.
DIMS System UX Development
Nobody chooses an internal system. They put up with it, and putting up with it is invisible.
Staff use the company's information systems because the job requires it, which means the friction never arrives as a complaint.
So the first task is finding where the work actually stalls, not asking what people dislike. Those are different lists and only one of them is worth fixing.
We ran research with the people doing the work, then took what it found into interface and graphics improvements.
PLM System UX Development
Product data outlasts the people who entered it, and the screen shows all of it at once.
A PLM system holds the development record for a product, which makes any single view crowded by nature.
On a dense screen the design decision is placement and route — where the things people look for every day sit, and how search and browsing divide the work between them.
Research ran alongside the interface and graphics improvements. It was carried out in parallel with the department's other systems so that the rules came out the same across them.
Maps System UX Consulting
Another internal system in the same department, and the same question: what to fix first.
Consulting here produces an order, not a design. Research-based, and delivered as a direction for improvement.
DMS — System Air Conditioner Central Management
Central management for system air conditioning, where a great many indoor units are watched at once and the graphics have to separate their states.
Central management for system air conditioning was moving to the web, where a manager works across a whole building at once.
Zones and floors are not a menu; they are how the person already thinks about the building. The navigation had to follow that, and the state of the equipment had to be readable across all of it without opening anything.
We took interface and graphics together and set the frame for the web screens. The later 3.0 work followed from it.
5 separate engagements on this line, 2014–2015.
Samsung Cloud Print
Graphics for the hosted cloud printing service, making the state of a print or a scan readable while it happens.
Connecting to a printer and storing to the cloud are two ideas the user should only have to hold as one. Interface design, developed alongside the graphics work.
3 separate engagements on this line, 2014–2015.
IoT Automation Service Design ToolKit UI/GUI Development
An automation rule is simple until the second condition arrives.
How conditions and actions are allowed to combine decides whether ordinary people can write a rule at all — the expressive version and the usable version are rarely the same. Interface and graphics for the toolkit.

Shinhan Mobile UI Standardisation Guidelines Restructuring for Design

Several mobile services from one company, each with its own idea of how a screen works.
Restructuring a standard is an argument about what is genuinely common — pull too much into the centre and every service has to fight the guideline to do its job. Built on the Shinhan mobile UX guidelines produced earlier.

Naval Combat System Seminar — CIC Multifunction Console UI/UX Research (Presentation)

The combat information centre (CIC) is the compartment of a warship where sensor information is compiled into a tactical picture and the ship's weapons are directed, with multifunction consoles as its operator workstations.
The combat information centre, called the operations room in some navies, is where a warship's sensor data is gathered, evaluated and turned into the tactical picture from which the ship is fought. Its operator workstations are multifunction consoles: general-purpose terminals of the combat system that can take on different warfare roles through software rather than dedicated hardware.
AEON prepared and presented a research paper on the CIC multifunction console at the naval combat system technology seminar hosted by LIG Nex1 in 2015.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
The combat information centre began in the Second World War as a room where radar and radio reports were plotted by hand on boards and tables. From the 1960s, computer-based tactical data systems replaced the plotters with consoles, and ever since the console has been the point at which the combat system meets the sailor. Its history is a good guide to how naval combat systems themselves have changed.
The first big shift was from bespoke military hardware to ruggedised commercial technology. By 2004 Lockheed Martin and DRS had delivered the 4,000th AN/UYQ-70, a COTS-based display and computing system fitted to submarines, Aegis cruisers and destroyers, aircraft carriers, amphibious ships and the E-2C. One family of workstations across so many platforms meant common spares, common training and a common upgrade path.
The Common Display System that followed took the idea further. For Aegis modernisation and DDG 1000, CDS brought high-definition displays on commodity hardware such as blade servers running Linux, and Aegis engineers studied how to use the extra screen space without overloading operators. Its consoles were designed for surface ships, submarines and aircraft and could run applications at different security classifications from a single workstation. The display was becoming a general-purpose window onto networked processing rather than a terminal tied to one computer.
The current shift is in software. In June 2026 Lockheed Martin delivered Aegis BL9.C3.0, the first baseline enabled by the Navy's Integrated Combat System, with a re-architected display component and a platform-as-a-service framework for containerised applications, on a planned six-month cycle of updates and certification. European suppliers have moved the same way, pairing new console hardware with combat management software that can be reconfigured for different roles.
The direction is clear enough. Consoles are turning into thin clients on a shared network; roles are assigned by software rather than by the physical position of a workstation; and large shared displays, and in some ships handheld devices, supplement the individual console. The CIC is becoming less a room full of specialised terminals and more a configurable space.
For operators, this raises the stakes of interface consistency. When the same console can be an air-warfare position one hour and a surface-warfare position the next, and software arrives every few months rather than every few years, sailors need interfaces that behave predictably across roles, ship classes and versions. Cognitive load, alert management and the transfer of skills between ships become design problems in their own right, not afterthoughts to hardware.

Naval Combat System Seminar — Open Architecture Combat System Using Mainstream COTS, Promotional Film (Exhibit)

An open-architecture combat system is a naval combat system built from modular components with standard interfaces, allowing mainstream commercial off-the-shelf (COTS) computing to take the place of bespoke military hardware.
An open-architecture combat system separates software from hardware through published, standard interfaces and middleware, so that individual components can be replaced or upgraded without redesigning the whole system. Building such a system from mainstream commercial off-the-shelf (COTS) computing, such as servers, networks and displays, lets navies refresh hardware on commercial cycles and reduces dependence on bespoke military-specification equipment.
AEON produced a promotional film on an open-architecture combat system using mainstream COTS components, exhibited at the naval combat system technology seminar hosted by LIG Nex1 in 2015.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
For most of the Cold War, naval combat systems ran proprietary software on military-specification computers that were designed once and then maintained for decades. Open architecture reversed that logic. By separating software from hardware through standard interfaces and middleware, navies can buy computing from the commercial market, refresh it on commercial cycles and upgrade functions without redesigning the whole system.
Aegis is the best-documented example. The US Navy modernises it through paired software and hardware upgrades known as Advanced Capability Builds and Technology Insertions; a 2016 contract for ACB 16 and TI 16 covered replacing combat system computer hardware, middleware and firmware on existing cruisers and destroyers, and noted that earlier ACB upgrades had already moved to commercial off-the-shelf computers.
Virtualisation was the next step. In 2019 the Aegis Virtual Twin, a prototype of a virtual Aegis combat management system, controlled a live missile engagement from USS Thomas Hudner while running in rugged cases small enough to fit under a dining table. By 2024 USS Winston S. Churchill was running an entirely virtual Aegis Combat System and used it to intercept an air target, with installation planned on further ships and land-based test sites. And in June 2026 Lockheed Martin delivered the first Aegis baseline enabled by the Navy's Integrated Combat System, with a framework for containerised applications and a planned six-month update cycle across several ship classes.
Korea's naval combat systems have followed a similar arc. Hanwha Systems describes its Naval Shield baselines moving from open architecture with COTS and MOTS technology in Baseline 2.0, introduced in 2008, to high-performance computing and virtualisation in Baseline 3.0 from 2017, with real-time middleware based on the OMG DDS standard, and lists combat systems for destroyers, frigates and submarines of the Republic of Korea Navy as well as frigates for export.
Open architecture is not free. Commercial components go obsolete quickly, so navies must manage a constant flow of hardware changes; shared software libraries need disciplined configuration control; and a more connected, software-defined combat system widens the cyber attack surface. The prize is that new sensors, weapons and functions can arrive in months rather than years, and that one software base can serve several classes of ship.
For operators, the faster cadence matters most. When combat system software changes every few months, the interface changes with it, and crews need consistency across versions as much as new features. Human-centred design has to move at the speed of software delivery, with operator feedback built into each release rather than gathered once at the end of a programme.

Naval Combat System Seminar — Next-Generation CIC Multifunction Console Design Consulting (Exhibit)

A CIC multifunction console is the general-purpose operator workstation of a warship's combat information centre.
The combat information centre (CIC) is the part of a warship where its sensor data is compiled into a tactical picture and its weapons are directed. The multifunction console is the standard operator workstation of the combat system there; a single console type can serve different warfare roles through software, which simplifies training, maintenance and upgrades across a ship and a fleet.
AEON carried out hardware design consulting for a next-generation CIC multifunction console and produced a 3D-printed model of the design, exhibited at the naval combat system technology seminar hosted by LIG Nex1 in 2015.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Warship operations rooms are being redesigned for the first time in a generation. The traditional model, rows of identical consoles each tied to a fixed warfare role, is giving way to spaces built around shared networks, reconfigurable positions and large collaborative displays, with artificial intelligence beginning to appear as an assistant to the operators.
Console makers have responded with new workstation concepts. Saab's Future Operator Workspace, first shown at DSEI 2023, combines ergonomic design with the company's 9LV combat management software and is intended to reduce mental load and fatigue and shorten reaction times; Saab developed it in about eight months with the University of South Australia and an Adelaide manufacturer. The emphasis on human-centred design, rather than on raw display size, is telling.
The Royal Navy's new frigates show two versions of the shared-infrastructure idea. On Type 26, a ship-wide digital backbone links common consoles and open-architecture software to server farms over a high-bandwidth network, and the INTeACT combat management system can be extended with downloadable software modules and reached from anywhere on board using handheld tablets. For Type 31, Thales's TACTICOS brings multi-purpose consoles that can be reconfigured for different operators, larger bulkhead displays for the command team, and rule-based monitoring that alerts operators only to contacts meeting defined criteria, while the commanding officer keeps weapon release authority.
Artificial intelligence has also been tested at sea. In 2021, during Exercise Formidable Shield, the Royal Navy trialled the Startle and Sycoiea applications aboard HMS Dragon and HMS Lancaster, the first giving live recommendations and alerts to sailors monitoring the air picture and the second evaluating threats and suggesting weapon assignments. Decision aids of this kind now feature on most combat management roadmaps.
Taken together, these programmes point to an operations room in which the console is a general-purpose access point, the tactical picture is shared on large displays, functions arrive as software updates, and machine assistants filter and prioritise information before it reaches a human. Pressure on crew numbers pushes in the same direction, since fewer people must handle more sensors and more simultaneous threats.
For operators, the challenge is coordination as much as individual workload. When positions can change role and much of the picture is shared, the team needs a common understanding of who is responsible for what and what the automation is doing. Interfaces that make machine recommendations transparent, keep alerting under control and support handover between people will matter more than any single piece of console hardware.

StyleMirror UI Development

A mirror that knows who is standing in front of it — studied years before that was an ordinary idea.
An early concept study for a smart mirror, at home and in a shop.
Everything shown on it competes with the reflection, which is the one thing the person actually came for. The research asked what an intelligent interface may put there, and where it may stand.

Home Trading System

Traders know where everything is. That is the constraint, not the opportunity.
On a system people use at speed and under money, familiarity is a feature you can destroy by improving. So the work was drawing the line between what to change and what to leave. Carried out jointly.
3 separate engagements on this line, 2013–2015.
K Project Accessibility Function End-User Usability Evaluation
An accessibility feature only shows its faults when someone who needs it tries to use it.
Who is recruited and what they are asked to do is the whole validity of the test. Evaluation carried out with the users the features were built for.
Samsung Hub UI(Android/Tizen Phone Type)
Music, video and books in one place, on two operating systems.
The service has to be recognisably one thing while sitting inside two platforms that disagree about how things are done. Phone-type interface designed for both.

S Health

Version 3.0 brought heart rate, coaching and much else. A first-time user opens it and has no idea where to begin.
A service used occasionally has one chance to show where things stand before it is closed again.
So the first screen is the whole service, and entering a record has to finish in fewer steps than the user expected. Anything longer and the next gap is permanent.
We took the issues raised by the previous version's usability work through into the 2.5 guide. Working with the same team meant the context did not have to be explained again.
4 separate engagements on this line, 2013–2014.
[Wingtip, Gear2] Wearable UX part Line-up UI
Small screens, short interactions, and several products in the line.
What the line-up shares has to be settled once, or each product answers the same question differently. A second phase continuing the interface work across the range.
Tizen Platform -based LCD Refrigerator of Food Manager Application
Used standing up, in a kitchen, with one hand, for a few seconds at a time.
A food manager on the refrigerator's own screen, opened in the middle of doing something else.
Registering an item has to cost less effort than remembering it would have — otherwise nobody registers anything and the rest of the app has no data to work with. Expiry dates had to announce themselves without being looked for.
We checked what the Tizen platform allowed and designed the interface within a short period.
Windows Phone 8.1 Applications
Interface and graphics for Samsung's apps on Windows Phone 8.1, balanced between the platform's design principles and the apps' own identity.
An app that ignores the platform looks foreign; an app that surrenders to it stops being yours. The graphics were made on that line. A second phase continuing the work.
3 separate engagements on this line in 2014.
Tizen for Simple UX of Commercialisation GUI Guideline Development
A simple phone experience, for people whose first smartphone this is.
Simplicity is subtraction, and subtraction is the argument — what goes, what stays, and who decides when the two collide at commercialisation. Written up as GUI guidelines.
Refrigerator LCD GUI — Export Markets
Canada means English and French on the same screen, and French is longer.
Mirroring a layout is not a translation task; it decides where the eye starts. Every relationship on the screen has to survive being reversed. We set out the conditions for each market first, then developed the LCD graphics.
2 separate engagements on this line, 2012–2014.
Sign Language Animator Usability Test Development
Whether signing reads correctly can only be judged by people who sign.
So the test had to measure comprehension, not preference — a hearing observer can tell you an animation looks smooth and nothing about whether it said anything. We designed the test and ran it.
Kids Platform
A kids mode has two users who must never meet on the same screen — the child using it and the parent setting its limits.
Kids mode shipped across several new handsets, and the graphics had to stay the same while the hardware and the platform underneath them did not.
So the child's screen and the parent's screen are separated before either is designed — everything else follows from where that line is drawn. Character work and graphic tone were settled alongside it.
Mass-production graphics and visual identity. 3 separate engagements on this line, 2013–2014.
Set Smart Rule Authoring Tool UX Development
People write the rules themselves, and rules get complicated in about two steps.
Authoring is a sequence problem: what you are asked for first decides whether the rest stays comprehensible. UX for the authoring tool.
Next TV Specialised UX Design and Working Prototype Development
An idea about television is hard to argue for. A working one is not.
So the question was which part of the experience had to actually run. We designed the specialised UX and built it far enough that people could operate it and decide.
Handset UI & GUI — Mass-Production Line-up
Forty-one handset and tablet programmes between 2008 and 2014, for a manufacturer shipping to carriers in several markets while the platform underneath changed three times.
Every model arrives with its own screen, its own keys, and an operator that wants something the last one did not. What is right for an open-market phone in Europe is wrong for the same handset at a US carrier, and wrong again in Russia or China.
At that rate the work is not designing screens. It is holding a line of them together while everything underneath moves. Feature phone to TouchWiz, TouchWiz to Android, Android to Bada and Tizen — and the base interface has to come through each of those recognisably itself.
So most of the effort goes into what gets decided once: the menu tree, the widget conventions, the base screens every model inherits, and the point at which a market’s requirements are allowed to change them.
Forty-one programmes over seven years, from model graphics through carrier and regional variants to mass production.
Store(Music/Books/Video/Learning/Game) UI
Five stores selling five unlike things.
A shared frame has to hold a song and a textbook without flattening either. Each store's interface was designed on the same structure.
2.4-inch Monochrome Display GUI
Two and a half inches, and no colour to tell anything apart.
Take colour away and contrast and shape have to carry every distinction — state, selection, menu and warning, all in black and white. Graphics developed for the monochrome display in a short period.
Zero Project for Accessibility Function End-User Usability Evaluation
Accessibility features in another new programme, put in front of the people they were built for.
Which features are tested and who is recruited decides what the evaluation can find. Run on the method established by the earlier evaluations.

Special Operations Support System Command & Control Screen Consulting

The special operations support system is a shipboard system that supports special operations conducted from the sea.
Maritime special operations are launched from ships or craft at sea, which carry the teams, boats and equipment and serve as their base during an operation. A shipboard special operations support system provides command and control for such operations, keeping the ship and the units it deploys connected.
AEON provided HCI consulting on the command and control display of the special operations support system.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Special operations from the sea depend on three things working together: fast craft, a parent platform, and the command and control that links them to each other and to forces ashore. In the United States, Naval Special Warfare boat teams operate the Combatant Craft Medium, a low-observable armoured boat used to insert and extract special operators in medium-threat environments. In August 2025 US Special Operations Command awarded ReconCraft an agreement to build an engineering development model of its successor, the CCM Mk 2.
The craft themselves are turning into strike and sensing nodes. SOCOM has tested a retractable eight-cell launcher on a CCM firing ALTIUS-700 loitering munitions, with the Mk 2 designed to carry such a launcher. Loitering munitions can scout ahead of a raid, provide fire support to boarding teams, or be handed off to forces ashore, which ties a small boat into a much wider targeting network. The US Navy has since sought industry information on loitering munitions for the craft under its Maritime Launched Effects Increment 1 effort.
Sensors and uncrewed partners are following. At SOF Week in May 2025, SOCOM's maritime programme office said it plans to fit forward-looking infrared sensors across its medium, heavy and assault craft and wants a fleet of autonomous uncrewed surface vessels by fiscal 2027, alongside interest in electronic warfare payloads. The war in the Black Sea, where uncrewed surface vessels have been used against a conventional navy, has sharpened attention on small, expendable craft that can scout, decoy or strike ahead of crewed boats.
The parent platform is changing too. The US Navy's expeditionary sea bases, such as USS Lewis B. Puller, are converted from a commercial tanker design and host helicopters, small boats and uncrewed vehicles, with upgrades planned to support special operations forces. A large, relatively inexpensive hull positioned offshore can act as a staging base, a maintenance hub and a command post for dispersed teams. The common thread across these programmes is connection: craft, drones, the parent ship and headquarters ashore need to share a picture over links that must stay hard to detect.
For operators, maritime special operations remain physically punishing. Small crews work at high speed, at night and in heavy seas, so information has to be readable at a glance and controls usable with gloves and limited attention. Adding loitering munitions and uncrewed vessels gives the same crews more to manage, while the staff aboard the parent ship must follow several dispersed teams over intermittent links. Interfaces that separate what needs action now from what can wait will determine whether these new capabilities help or overload the people using them.

Submarine Combat System UX Consulting

The Jang Bogo class comprises nine diesel-electric submarines of the German Type 209 design in service with the Republic of Korea Navy.
The Jang Bogo-class (Type 209) submarines were delivered to the Republic of Korea Navy between 1993 and 2001, with all but one built in South Korea. A capability upgrade programme has modernised the boats' combat system, sonar and periscope, taking the combat system of the Navy's later Type 214 submarines as its base.
AEON provided HCI consulting for the combat system interface of the upgraded submarines.
Technology trends
Written by AEON from public sources. It describes the field, not the project above.
Submarines routinely outlive their electronics several times over, so the combat system upgrade has become a normal stage in a boat's life rather than an exception. South Korea's Jang Bogo-class boats show the pattern: ROKS Jung Woon returned to service in August 2024 with a locally built combat system, an improved periscope and a new towed sonar, after modernisation work by Hanwha Ocean, extending the useful life of a hull built in the late 1990s.
The US Navy turned this into a continuous process with AN/BYG-1, the combat control system used across its attack and ballistic-missile submarines and in Australia's Collins class. Rather than waiting for long refits, the programme refreshes hardware with commercial processors through periodic technology insertions and delivers software in biennial advanced processor builds, with the tactical control and weapons control elements developed in parallel. Commercial hardware and a regular cadence keep processing power close to the state of the art and let new sensors and algorithms reach the fleet in years rather than decades.
European builders have moved the same way. Germany and Norway's Type 212CD boats will carry ORCCA, a combat system from kta naval systems, a joint venture of TKMS, Atlas Elektronik and Kongsberg; in December 2025 Kongsberg signed for its share of the system for six boats, four German and two Norwegian. When it was unveiled, ORCCA was presented as modular, able to take subsystems from customer countries and to bring data from many onboard systems to one multifunction console. For export customers that modularity matters, because navies want to integrate national sensors and weapons and keep control of upgrades.
The next step is in signal processing. The AUKUS partners have pooled sonar data from their P-8 aircraft in a shared environment to train AI tools that filter out biological noise such as whales and shrimp, so that operators can concentrate on man-made sounds. The same reporting noted that the US already uses a similar tool aboard an attack submarine and that the partners intend eventually to extend these tools to submarines and surface ships. Machine learning is being aimed first at the most labour-intensive part of undersea warfare: sorting genuine contacts from a noisy ocean.
For crews, these trends change the nature of the work. Sonar operators and fire-control parties will increasingly receive machine-generated cues and classifications, and their trust depends on seeing why a contact was flagged, how confident the system is, and how to check it against the underlying data. Frequent software builds also mean that interfaces evolve during a crew's time aboard, so consistency between versions and clear explanation of what has changed become part of keeping a submarine ready for patrol.
— AEON has carried out more than 800 UX projects since 1998. —
What is shown here is a selection.