ComBoard

Timeline:
Apr 2024 – Jun 2026
Role:
UX Research, Design & iOS Development (solo)
Tools:
Figma, React Native / Expo, Swift

The AAC app that took one student from spelling twice a month to five times a week — now a shipped iOS app for any non-speaking communicator.

App Store → Figma Prototype →
Animated screen recording showing the ComBoard prototype in use on an iPad

Overview

ComBoard is a custom AAC (augmentative and alternative communication) app designed for a 16-year-old non-speaking student with cerebral palsy. Her existing tools — a Tobii eye-tracking device and a plexiglass E-TRAN board — were bulky, slow to set up, and blocked her face during use, creating barriers to both communication and social connection.

How might we design an AAC app that minimizes eye strain, integrates predictive text, and adapts to the individual — so she can communicate independently without obscuring her face?

Goal #1

Increase spelling-to-communicate from 2x/month to 4x/month

Goal #2

User preference rating at least 20% higher than existing methods

Meet Ellie

Photo representing Ellie, the user research participant

Ellie is 16. She communicates through left-right eye movements with interpretation support from an adult aide. The tools available to her — a Tobii device and a plexiglass board — cause eye strain, block her face, and make every interaction feel mediated rather than spontaneous.

Results at a glance

9.8× communication usagevs. prior tools (ETRAN, Tobii)
5/5 user preference1/5 (Tobii) and 3/5 (plexiglass board)

Ellie's Goals

  • Spontaneous communication
  • Enriched social interaction
  • Active class participation
  • Reduce physical strain

Ellie's Frustrations

  • Bulky, expensive solutions
  • Calibration lag for eye-tracking
  • Slow setup disrupts natural flow
  • Limited predictive vocabulary

Journey Map: Communicate a Thought

Ellie wants to communicate independently and in real time. Here's what that looks like today.

1. Trigger — Need to Communicate
Tasks
Feels the need to ask a question or share an idea; signals with eye movements or gesture
Feeling
Urgent, Anxious
Improvement
Always available AAC that allows self initiation
2. Wait to be Noticed
Tasks
Wait quietly; vocal noises to get attention
Feeling
Helpless, Dependent
Improvement
Visual or audio cues built into app to alert when she has something to say
3. Interpretation Attempt
Tasks
Aide interprets intent through body language and eye gaze; possibly set up bulky Tobii or plexiglass E-TRAN
Feeling
Embarrassed, Impatient
Improvement
Smart, predictive text
4. Classroom Participation
Tasks
Tobii blocks Ellie's face, E-TRAN is unwieldy; limited to aide-proposed options
Feeling
Frustrated, Excluded
Improvement
Eye-tracking technology; use of common device screens that don't block face

Accessibility

These decisions came from recurring user interviews and a competitive audit of existing AAC tools.

User Constraints

  • Communication boards have limited word categories and lack spelling options for free expression.
  • Many devices are cluttered and require fine motor control, causing eye strain.
  • Many devices limit social interaction by blocking the user's face and view.

Design Decisions

Create a keyboard that integrates predictive text.

Cluster letters and add phoneme keys (e.g. "TH") to reduce small eye movements and mis-triggered selections, lowering motor demands.

Introduce an animated "clear" state that removes unused keys after each selection to reduce clutter and allow eyes to rest.

Design for standard tablets and phones so communication partners can hold the device, preserving eye contact and social connection.

Keep visually minimalistic with limited colors, strong contrast, and clear grouping to reduce cognitive load.

Design Decisions

I prioritized getting a functional keyboard into Ellie's hands as fast as possible, deferring polish and settings in favor of early, frequent testing with the actual user.

Ideation & Drafting (Week 1)

Draft 1 was a category-based communication board on Google Slides — a common format, but not scalable and limited to pre-set options. I switched to a keyboard design that puts more power in the user's hands.

Early communication board design iteration showing category-based layout Second iteration of the communication board moving toward a keyboard layout

Custom Designing Icons (Week 2)

After learning Ellie struggles with spelling, I added small icons to each key as visual anchors. I sketched on paper, built them in Figma, and tested layouts with Ellie to balance letters, icons, and white space.

Paper sketches of keyboard icons alongside early digital icon components in Figma

UI (Weeks 3–7)

Figma design file showing paper sketches and early digital component iterations for the ComBoard keyboard

Interaction, User Testing, Prototyping (Weeks 8–12)

ComBoard prototype showing the interaction flow and user testing setup

The Work

Figma Make Iteration

When Figma Make launched in July 2025, I revisited the prototype to add predictive text, text-to-speech, and speech-to-text — and to push personalization further for a broader range of AAC users.

  • Refined the UI to make better use of screen space
  • Switched from custom to Lucide icons for consistency
  • Improved selection feedback with scaling and highlighting
Refined ComBoard UI from the Make-a-Thon showing improved screen space usage ComBoard settings and personalization options from the Make-a-Thon prototype ComBoard Make-a-Thon prototype showing selection feedback improvements
ComBoard Make-a-Thon iteration with Lucide icons and updated layout
View Figma Prototype →

Sticker Sheet

All components organized by variant and state for easy iteration and developer handoff.

The App

In June 2026 I took ComBoard from Figma prototype to a shipped native iOS app in a one-month sprint — handling design, development, and App Store release solo.

Tech Stack

  • React Native 0.81 + Expo SDK 54 (TypeScript) — fully on-device, no backend
  • Native Swift module (StoreKit 2) for in-app purchase
  • Apple system Eye Tracking (iOS 18+) for hands-free use
  • On-device text-to-speech + speech-to-text, Lucide icons, React Navigation
  • iPhone + iPad, iOS 16+

Key Features

14 color themes including colorblind-safe (Paul Tol) palettes and monochrome sets

9 keyboard layouts (2-cluster to 12-cluster) with frequency-based letter ordering and range-collapsing tiles

Apple Eye Tracking — a pivot from the planned WebGazer research to native iOS integration

Text-to-speech + dictation — auto-speaks the message on pause, with speech-to-text input for communication partners

Emoji keyboard with the same recursive cluster drilling interaction

Adaptive sizing across phone and iPad, portrait and landscape, with range-collapsing cluster tiles ("A–R")

Shipped Product

ComBoard shipped app showing the 6-cluster keyboard with color-coded letter groups, phoneme keys (TH, SH, CH), prediction bar, and Yes/No buttons
ComBoard 2-cluster layout showing 'I WANT SOME' typed with contextual word predictions: HELP, WATER, WANT, FOOD, YOU
ComBoard settings panel showing 13 color themes including Classic, Pastels, Natural, Dark, two colorblind-safe palettes, seven monochrome sets, and layout selection
View on App Store →

Outcomes

A line graph showing device usage per month from April 2024 to June 2025. The student requested to use the Tobii device the least with a nearly horizontal line at zero, plexiglass board occasionally, and the digital ComBoard most often with a peak of 24 uses in January 2025. Plexiglass board usage decreased as digital board usage increased. There is a dip in ComBoard use during December 2024.

Goal #1 Result

9.8× increase in communication usage vs. her prior tools (ETRAN, Tobii) — from 2x/month to 5x/week with ComBoard.

Goal #2 Result

Method preferences: Tobii (1/5), Plexiglass board (3/5), ComBoard (5/5)

Shipped

ComBoard is now on the App Store — available to any non-speaking communicator. On-device, private, no subscription.

Reflection

What I Learned

  • Use existing resources. Custom icons delayed the product and introduced inconsistencies — Lucide already had what I needed.
  • Color is functional, not just aesthetic. The palette went through several iterations before landing on colors that help Ellie focus on selections by looking toward color regions rather than distinguishing individual letters.
  • State management matters. Complex interaction states need careful planning — poor implementation leads to messy code and a frustrating experience.

Next Steps

Refine Eye-Gaze & Switch Support

Apple system Eye Tracking replaced the planned WebGazer research — next, refine dwell and gaze UX and ensure compatibility with assistive tech hardware like switches.

Broader User Research

Gather feedback from additional AAC users beyond Ellie to validate design decisions and guide the next round of iteration.

Custom Word Banks

Add the ability for users and caregivers to manually add common words and phrases, making prediction more personal and useful from day one.

Android

Bring ComBoard to Android via React Native to reach more non-speaking communicators.