Accessibility testing
AI accessibility testing by people with accessibility needs
Real people with accessibility needs test the AI you already run, or are about to launch: screen reader users, Deaf and hard of hearing customers on text channels, people with speech differences on voice lines, and people for whom long or complex replies are a barrier.
Usually a fit
Custom Panels
Quoted
Larger or more specific panels for launches, languages and accessibility.
- Larger panels, from 15 to 50+ testers
- Specific personas you define: age groups, accents, first-time users, frustrated repeat callers
- Pre-launch panels for an AI that is not live yet
- Multilingual and accessibility panels
AI accessibility testing asks whether the people most likely to depend on a self service channel can actually use it. A chat widget may look fine on screen and still trap a screen reader user in an unlabeled frame. A voice agent may work well for most callers and repeatedly misunderstand someone with a stutter or a speech difference. A reply that is correct but long and dense can be a real barrier for someone with a cognitive disability.
Automated accessibility scanners check code against rules, and they are worth running. They cannot tell you whether a blind customer was able to find the chat, follow the conversation and finish a return, or whether a caller with a speech difference was understood without being asked to repeat themselves five times.
Our panels include testers who live with these needs every day and use their own assistive technology. They carry out real customer tasks and score each session on the same fixed rubric, with written notes about every barrier they hit.
To be clear about scope: this is not a WCAG audit, an ADA compliance review or a legal certification, and we do not issue one. It is a human view of whether your AI works for people with accessibility needs, which you can use alongside any formal review your team arranges.
Signals
When it is time
If any of these sound familiar, a panel will tell you what you need to know.
Your AI replaces a phone line or a person
When a human channel is reduced, customers with accessibility needs are often the ones most affected.
You serve older customers or public services
Healthcare, utilities, government, banking and insurance customers include many people who rely on assistive technology.
Your chat widget came from a third party
Embedded chat widgets vary widely in how they work with screen readers and keyboards, and they are easy to overlook.
Your voice agent relies on speech recognition
Callers with speech differences, hearing loss or slower speech need a voice agent that waits, confirms and offers alternatives.
You want real feedback alongside automated scans
Human testers show how the barriers an automated scanner flags, and the ones it cannot see, feel in an actual conversation.
How the panel is set up
How it works for this use case
The panel is built around this moment, step by step.
- Step 1
Agree the needs and channels
We choose the accessibility needs and channels that matter to your customers, such as screen readers on web chat or speech differences on voice.
- Step 2
Match testers with lived experience
Testers use their own assistive technology and settings, the way they do every day.
- Step 3
Run real customer tasks
Each tester completes realistic tasks like checking an order, booking an appointment or asking for a person.
- Step 4
Score and describe every barrier
Sessions are scored on the fixed rubric, and every barrier is described in plain words with the moment it happened.
Deliverables
What you receive
Everything you need to decide what happens next.
- Rubric scores for each session, grouped by accessibility need
- A plain description of every barrier testers hit, with where it happened
- Transcripts or recordings of each session, with tester notes
- What worked well and should stay as it is
- A one-page verdict: keep it, fix these settings, or reconsider it
We only diagnose. What you do with the findings, and who does it, stays entirely your call.
Questions
Common questions
Quick answers before you request a test.
Is this a WCAG or ADA compliance audit?
No. We do not certify compliance with WCAG, the ADA or any other standard, and this is not legal advice. It is human testing by people with accessibility needs, which complements a formal audit rather than replacing one.
Which accessibility needs can testers cover?
Common panels include screen reader users, keyboard only users, Deaf and hard of hearing customers on text channels, people with speech differences on voice channels and people who find long or complex replies difficult. We confirm the mix when we scope the panel.
Do testers use their own assistive technology?
Yes. Testers use their own devices, screen readers and settings, because that is how your customers will reach your AI.
Keep reading
Related AI types and guides
More on the AI types this applies to.
Support Chatbots
Real people chat with your support bot as angry, confused and refund-seeking customers.
Read more about Support ChatbotsWhat we testAI Voice Agents
Real callers with real accents and impatience test your AI phone and voice agent.
Read more about AI Voice AgentsWhat we testAI IVR
Human callers test conversational and AI IVR menus end to end.
Read more about AI IVRWhat we testBooking Assistants
Real people book, reschedule and cancel with your scheduling AI.
Read more about Booking AssistantsGuideHow to Test a Chatbot With Real People
Scope, personas, scenarios, scoring and reading the results, step by step.
Read more about How to Test a Chatbot With Real PeopleGuideHow to Test a Voice Agent Before Go-Live
Real callers, accents, interruptions, noise and handoff.
Read more about How to Test a Voice Agent Before Go-LiveGuideThe Chatbot Testing Checklist
A printable list of scenarios and personas for any chatbot.
Read more about The Chatbot Testing ChecklistSee your AI the way your customers do
Tell us what your AI does and what worries you. We build a panel of real people around it and hand you every transcript, a score per criterion and a plain verdict. We never sell the fix.