AAC / Wren
Faster conversations, in your own words
An iPad app for people who type to talk, with suggestions that speed them up without putting words in their mouth.
- Role
- Product Designer
- Solo project
- Timeline
- October 2026
- Research to prototype
- Type
- iPad app
- AAC
- Focus
- Accessibility
- AI authorship
Overview
People with ALS, cerebral palsy, or aphasia often talk by typing on a tablet that speaks for them. It’s slow, and AI prediction could close much of the gap. But a fluent sentence someone didn’t quite mean, said in their voice, is worse than a slow one they did. Wren is a working iPad prototype built around that tension: the AI suggests, the person decides, and every word they send stays recognizably theirs.
8–10words a minute typing to talk, against 130–200 in speech
63–72%fewer taps with word suggestions
13walkthrough findings, all addressed
0words sent off the iPad, voice included
TL;DR
- The problem
- People who type to talk speak at about a tenth of the pace around them, and the AI that could close the gap risks putting words in their mouths.
- What I did
- Synthesized published AAC and AI research into seven design rules, built a working iPad prototype with an on-device voice, then ran a walkthrough with three users and their SLPs, and addressed every finding.
- The outcome
- Suggestions that cut typing effort by about two-thirds in tests, where every suggested word can be seen, swapped, or refused.
Sorry! The rest of this case study is password protected.
For recruiters & hiring managers: email me and I’ll send the password.