Big changes are happening in how technology companies make their websites and apps accessible to everyone. This weekly update focuses on new AI agents designed specifically to help people with disabilities use digital products more easily.

Level Access, a leading company in digital accessibility based in Stafford, Virginia, USA, launched powerful new AI tools on October 6, 2025. These tools act like smart assistants that work alongside human experts to find accessibility problems faster, help developers fix them more quickly, and prove to company leaders that accessibility efforts are working. The company built these AI agents because development teams face increasing pressure to build products quickly while also making sure everyone can use them, including the 1.3 billion people worldwide who have disabilities.

The CEO of Level Access, Mark Zablan, explained that embedding AI agents throughout their platform helps lighten the workload for people working hard every day to make the internet more accessible. The AI agents fit right into the tools and processes teams already use, like their development workflows and reporting systems, so workers don't need to learn completely new ways of working.

These new tools address three major problems that slow down accessibility work. The first problem is that teams often can't understand their real risk of failing accessibility rules because they're drowning in too much repetitive information stored in separate documents. Imagine trying to find specific problems when you have hundreds of reports scattered across different files - it's nearly impossible to figure out what matters most.

The second problem is that developers can't fix issues fast enough because they lack clear, specific guidance about what to do right when they need it. Without good instructions in the moment, developers might fix problems the wrong way or waste time searching for answers. The third problem is that teams struggle to show their progress in ways that convince executives to keep funding accessibility programs, which sometimes causes these important programs to shut down.

Level Access organized their new AI features into three categories that match these challenges. The "Find" features include tools that group the same problems together so teams can fix multiple issues at once instead of one by one. They also have AI filtering that lets people use everyday language to ask questions and get the most important problems highlighted. Monitoring summaries show what changed between checks and suggest what to do next.

The "Fix" features offer several helpful capabilities. Code Suggestions gives developers specific guidance on how to repair problems right in their workflow. Audit Summaries take manual evaluations and turn them into action plans. Ask Level AI works like a knowledgeable assistant trained on accessibility rules and 25 years of industry experience from Level Access experts. There's even a Model Context Protocol Server that helps AI assistants work better with the Level Access Platform so developers can fix things with fewer interruptions.

The "Prove" features include a Reporting Agent that quickly creates summaries for executives, highlighting both what got fixed and what still needs attention. These reports help leaders understand the value of accessibility work and make decisions about continuing to fund these efforts.

Andrew Chung, Chief Product Officer at Level Access, emphasized that these agents aren't just smart - they work directly where teams already do their jobs. This approach makes accessibility a natural part of how organizations operate rather than an extra burden that changes everything about how people work.

The company uses what they call "hybrid intelligence," which combines automated platform features, AI agents, and oversight from human experts. This combination ensures the system remains trustworthy and meets enterprise-level compliance requirements. Alistair Garrison, Head of AI Innovation at Level Access, explained that their approach helps shift accessibility from something teams check occasionally to a process that's built in from the start.

Experts in the accessibility field also gathered for a webinar to discuss both the promise and challenges of AI in accessibility. They talked about the future of "agentic AI" that could understand each person's specific abilities and create custom interfaces on the spot. This could mean someone could book a flight by voice, typing, signing, or exploring an interactive map - whatever works best for them.

However, these same experts raised important concerns. Joe Devon, founder of A11y Audits, pointed out serious security risks where powerful AI systems with broad access could be tricked into stealing sensitive user data like credit card numbers. There's also the human factor - if companies still struggle to accommodate different disabilities when hiring real people, how can AI be expected to do better when it learns from us? This concern grows more urgent as more companies use AI to screen job applicants.

On a positive note, embedding AI into everyday products like Microsoft 365 and social media platforms will dramatically reduce the number of images without proper descriptions for people who can't see them. Most of Level Access's new capabilities are available now, with some features still in testing or rolling out later this quarter.