AI-Agent Readiness & Human-Verified Assurance

Prepare for what comes next. We help organisations evaluate the accessibility and usability of AI-powered features, investigate the limitations of simulated user testing and explore how emerging AI agents interact with digital services.

Combining expert manual testing, assistive technology evaluation and, where appropriate, research with disabled and neurodivergent participants, we identify barriers that automated approaches may overlook. Our work connects emerging AI technologies with real-world accessibility, usability and human oversight.

Discuss AI accessibility assurance

AI Accessibility Across Three Areas

AI-Powered Product Features

Accessibility and usability testing of AI features introduced into websites, applications and digital services, from search and recommendations to generative experiences and conversational interfaces.

Human-Verified AI Research

Critical evaluation of AI personas, synthetic users, simulated journeys and AI-assisted research methods to identify misleading assumptions and unexpected barriers.

AI-Agent Interaction Readiness

Exploratory testing of how emerging AI agents interpret, navigate and interact with digital interfaces and changing content states.

When AI Changes the Interface

Introducing AI functionality can create new accessibility and usability challenges even when the surrounding product is otherwise accessible. We test whether AI-powered features remain understandable, operable and usable through different interaction methods.

That includes screen reader accessibility, keyboard and voice control, dynamic content and focus management, clarity and comprehension, error handling and recovery, transparency and user control, and mobile accessibility. Testing uses relevant tools such as JAWS, NVDA, VoiceOver, TalkBack, Apple Voice Control, Windows Voice Access, magnification and keyboard navigation, with the test setup selected according to the project scope.

Real Discovery Cannot Be Prompted in Advance

Synthetic user research and AI-generated personas can reproduce the assumptions embedded in their prompts. If a simulation is told in advance what barrier to expect, it can validate the original assumption rather than reveal unexpected issues.

Human research can uncover previously unidentified accessibility barriers, unexpected interaction strategies, workarounds, cognitive friction, differences between apparent and actual task completion, and barriers affecting independence. AI-assisted methods may be useful, but synthetic personas are not substitutes for disabled people’s lived experience. Human verification helps teams understand the strengths and limitations of AI-generated findings.

Preparing Digital Journeys for Emerging AI Agents

Some interface characteristics that support human accessibility can also help selected AI agents interpret and navigate digital experiences. Semantic HTML, clear headings, accessible names, predictable focus, descriptive links and buttons, robust forms, meaningful error messages and consistent state changes are all important.

Where commissioned, we can evaluate how selected external AI agents interact with defined journeys. Agent behaviour varies by technology and interaction method, so this is exploratory rather than universal compatibility testing. WCAG conformance does not guarantee AI-agent compatibility, and successful AI-agent navigation does not establish accessibility for disabled users.

Human Expertise Across Emerging Technology

Our approach is shaped by the feature, the user goals, the risks and the relevant accessibility requirements. We begin by understanding the journey, the likely failure points and the environments in which it will be used.

We then evaluate real interactions through expert manual testing, relevant assistive technologies and, where included in scope, research with disabled and neurodivergent participants. Where AI-generated outputs, personas or external agent behaviour are relevant, we assess them critically and translate findings into structured observations, practical recommendations and follow-up validation guidance.

Evidence for Accessible AI Experiences

Depending on scope, outputs may include AI feature accessibility findings, manual assistive technology observations, AI search and conversational interface evaluations, human-verified AI research analysis, AI-agent interaction observations, user impact evidence, supporting screenshots or recordings and prioritised recommendations. Findings distinguish verified accessibility barriers from exploratory AI-agent observations and broader AI usability concerns, while making clear which issues require product changes and which are worth monitoring as AI capability evolves.

Inclusive Research & Disabled-User Testing

Research with disabled and neurodivergent participants. Explore Inclusive Research.

Assistive Technology Health Checks

Expert manual testing of digital experiences across assistive technologies. Explore Health Checks.

Inclusive Design & User-Centred Product Development

Accessible product design and development support. Explore Inclusive Design.

Granular Journey Discovery & Moderated Research

Structured evaluation of digital journeys. Explore Journey Discovery.

Accessibility Audits & Action Roadmaps

Technical evaluation and remediation planning. Explore Accessibility Audits.