I turn complex digital products into clear, scalable workflows, bringing a strong focus on digital accessibility. Experienced in end-to-end product design, redesign projects, user research, and UX strategy.
Aktivtisch
Product design for touch, made accessible for older adults.
Aktivtisch is a touchscreen activity table for care homes and disability services. I am its only product designer. I ran the research, redesigned settings and nine game apps, and built the design system.
UX/UI
Accessibility
Research
Design System
Product Management
Overview
Objectives
- Learn why caregivers avoided settings.
- Make settings, profiles and shutdown safe for untrained staff.
- Bring nine game apps to one accessible standard.
- Give the developers specs that explain themselves.
Role: Senior Product Designer, the only product designer
Period: Aug 2025 – present
Team: Stakeholders · Unity developers · content designer · QA
Key skills: Mixed-methods research, Prioritisation, Accessibility for older adults and tremor, Touch and remote interaction design, Design systems, Prototyping and on-device testing, Design QA, Art direction
Goal
Sales. Build an accessible, research-backed interface that gives care homes and investors a reason to buy.
Standing apart. Profiles set Aktivtisch apart from competing tables. Make them easy enough for daily use.
Consistency and scale. Build every screen and new app from one design system, so releases cost less.
Daily operation. Cut the time staff spend on setup, troubleshooting and navigation.
Engagement. Make sessions longer and more enjoyable with clearer type and calmer backgrounds.
Context
Three table sizes and a tablet
The table comes in 43, 49 and 55 inch. Each one tilts into a table, lectern or board position. Aktivtisch GO runs the same apps on a tablet. Every layout I designed had to work on all of them.
One Unity app for Windows and Android
Residents watch from up to three metres away, so I kept layouts simple and legible. The table takes touch and a remote cursor. The 43-inch table runs Full HD; the 49 and 55 inch tables run 4K.
A Unity app has no system controls. I designed and specified a full library of accessible custom components, down to the keyboard. Each has a 64 × 64 px minimum touch target and Atkinson Hyperlegible type.
Discover and define
Benchmarking Aktivtisch against four competitors
Buyers compare activity tables, so I benchmarked four competitors on format, training, support and features. Aktivtisch was the only table in three sizes and one of two with profiles.
Caregivers run the table
Research showed staff do more than 95% of the navigation. Residents tap the games. So I designed for two users at once.
The table serves residents, caregivers who run sessions, caregivers who manage the table, and facility management. Personas are an old-school tool, but my proto-personas aligned stakeholders on these four groups. With my journey maps, they surfaced the pain points we prioritised. One was "I have to hold his hand just to lower the sound."
Three problems shaped the redesign
Settings nobody trusted. I designed a survey for our client facilities. 29 of the 308 invited answered. All found launching an activity easy, yet 55% felt unsure in settings. 59% rarely or never created profiles. Their comments named problems in the most-used features, like saving changes and adding web pages.
Games that punished mistakes. In a heuristic evaluation of ten apps, I coded 105 notes against an 18-question checklist. In three apps, players got no second try after a wrong answer.
A screen too big to reach. On two field visits in disability care, residents could not reach the top corners. They expected a tap to work the moment their finger landed.
The root cause was a mismatch. The system thought in profiles and categories; caregivers thought "open an app, run the activity, done".
Prioritise
Ranking 123 ideas before building any
I built a MoSCoW framework with three lists to rank 123 ideas for the beta. Each idea's score is its value divided by build effort. Each idea also gets a development level from 1 to 3. Level 1 is daily use and gets the most care; level 3 only has to work.
I ran a MoSCoW workshop with stakeholders and the developers. Nothing started as a Must-have. 38% to 48% of each list went to Won't have for the beta. Stakeholders then adopted the ranking as the team's way to prioritise.
Design
Touch targets for tremor, built from scratch
I found no design system for people with tremor or dementia on a 55-inch screen. I turned 47 age-related changes into rules, then components. A target shows the press when a finger lands. The answer counts when it lifts. The table's built-in tremor filter tells a shaking hand from a real tap.
Every control also works with this remote's cursor.
How it was tested
- Two field visits in disability care
- White headlines fell below 3:1 contrast on 9 of 21 pictures
Settings on one screen, in three groups
Full-size prints on the real 55-inch screen showed my dialogs 20–30% too large, so I shrank them.
Wizards for albums, web pages and the logo
I turned each risky flow into a short wizard that asks one thing per screen. All three share one file browser, so caregivers learn it once.
Shut down
Caregivers unplugged the table because shutdown was tucked away in the settings and needed a long press. I made shutting down easy to find and turned it into a proper PC shutdown. If someone unplugs the table anyway, the next start shows how to shut down safely.
Profiles in one short form
Facilities named profiles after sessions twice as often as after residents. So I made the form's example name a session, "Afternoon group".
Consent and product keys
German law requires Decline to be as easy as Allow, so I gave both equal weight. I renamed licence keys to product keys. Their errors show under the field.
A keyboard built for this screen
Typing is the slowest input on the table. I designed a custom keyboard with German, English, Chinese and Hong Kong English layouts.
Six rules every game must pass
In Test Your Knowledge, Guess the Saying and Black Sheep, a wrong answer was final. From the review, I wrote six rules for every game. Each game keeps its mechanic, and every round ends. No wrong answer is final, and difficulty lives in the content. Everything reads at three metres. Every main action has text and an icon.
Consistent backgrounds with a human touch
The review found busy backgrounds that hid the picture, and gloomy dark ones. Every redesigned game now sits on one family of light backgrounds in the brand greens. Hand-made SVG illustrations stay sharp on every table size. Each game gets a small motif at the edges, so the middle stays clear.
Game app redesign example: Black Sheep
Caregivers did not know the three apps were levels of one game. I merged them into one tile, entered by topic, with levels inside. I set difficulty by how much of the picture changes. A soap dish at 18% is Beginner; an umbrella at 0.54% is Expert. I audited 83 library photos against that spec and rejected 45.
Saving a painting with one field
I removed the author field from the save dialog. With no name typed, the table names the painting after the profile, like "Anna's painting".
One set of media controls for songs, stories and videos
Singing, the fairy tales and Advertising Classics each placed their buttons differently. Now they share the same buttons, sizes and states. Lyrics follow the song and can be hidden to show the whole picture.
Test
Testing on the table itself
I coded prototypes for nine apps and made 25 offline test builds. On the table, a one-second long-press that worked with a mouse failed under a real finger. The builds also needed true full screen and an exit button for touch. Slideshow text and controls match the design within one pixel.
Deliver
A design system the developers can build from
In Figma, I built the Master File with about 1,600 components and variants. In Confluence, I wrote the UI guide on how the interface behaves and why. Every screen uses a 16-column grid with a 24 px safe area. The developers build each new app from tested components. The aim is a consistent product that scales, with less mess and cheaper releases.
Localisation in three languages
The table ships in German, English and Chinese. German runs about 40% longer than English, so I gave every text slot a length limit. I wrote notes with a character counter for the content designer. All 1,291 imported German questions fit their slots.
AI for prototypes, then design by hand
I used AI only to build coded prototypes fast. Each one showed the developers how a screen should behave. Then I revised every design by hand in Figma.
Animation, all custom. I designed every animation to a few timing rules. Screens cross-fade in 350 to 450 ms, and rewards get more time. The Black Sheep star bursts over the odd picture and is gone within 1.8 seconds.
Brand
One brand and customer experience
I treated the brand as the whole customer experience. A care home meets Aktivtisch online and in print, then uses it every day. I art-directed the brand book and the January 2026 photo shoot with real residents. I also designed the new website concept. The logo set and flyers were made with an external designer. I chose Commissioner and Source Sans 3 for the brand, and Atkinson Hyperlegible inside the product.
Measure
Measuring what keeps care homes using the table
Content that never changes wears residents out. I proposed smaller, more frequent updates of about 1 GB each, and stakeholders made them monthly. Each update is a retention hook, and NEW badges on Home point caregivers to the fresh content.
Regulations make care homes hard to visit, so every table reports its daily use. Counting opens of 15 seconds or more filters out accidental taps. Tables that stay offline store each day and send it later. I designed the consent with Decline as prominent as Allow.
What I learned
Prioritise with one shared score.
I gave all 123 ideas one score, value divided by effort. In the MoSCoW workshop with stakeholders and the developers, nothing started as a Must-have.
Ship the necessary first.
Settings and shutdown went first, because the survey showed caregivers struggled there every day.
Cut what does not pay its way.
A full redesign of the old apps, about six months of work, became light fixes.
Keep stakeholders aligned.
Every decision went into Jira, so the team worked from one record. Sales joined early and used the survey to show that staff unplugged the table. That finding made shutdown easy to find.
Next case
Mercedes me App
My unusual user research that proved useful for Mercedes-Benz
