Accessibility’s Next Chapter: How Coding Assistants, adaptive overlays, and agentive AI personas will reshape Inclusivity

Web accessibility has come a long way since the introduction of foundational standards like WCAG, ATAG, and UAAG. These guidelines provided the building blocks for creating interoperable and inclusive digital experiences, offering a shared framework for web content, authoring tools, and user agents. Beyond technical compliance, they also served as tools for advocacy and education, helping developers understand how people with disabilities use the web. Yet, for all their progress, accessibility remains far from a default practice in modern design and development workflows.

In 2019, Laura Johnson ’s article Making the Web More Accessible Using Machine Learning explored how artificial intelligence could transform accessibility. Her work highlighted the importance of collaboration, annotated datasets, and automation in addressing accessibility barriers. While these ideas were prescient, the rapid evolution of AI since then—particularly generative AI—has introduced new opportunities and challenges. Tools like GitHub Copilot and Cursor enable developers to embed accessibility into their workflows, while overlays and testing tools attempt to make compliance faster and easier. However, each approach raises critical questions about effectiveness, user needs, and the balance between automation and human insight.

This article builds on Johnson’s foundation, examining the challenges of today and envisioning a future where generative AI and adaptive solutions make accessibility seamless, proactive, and inclusive. By combining technological innovation with the empathy of user-centered design, we can create a digital world that truly works for everyone.

Foundations of Accessibility

The late 1990s and early 2000s marked a pivotal era for digital accessibility. During this time, key guidelines emerged from the W3C that provided the foundation for creating more inclusive web experiences:

• Web Content Accessibility Guidelines (WCAG): Introduced in 1999, WCAG established principles to ensure web content was perceivable, operable, understandable, and robust for users with disabilities. These guidelines became the cornerstone of accessibility efforts.
• Authoring Tool Accessibility Guidelines (ATAG): Published in 2000, ATAG focused on accessibility within the tools developers use to create web content.
• User Agent Accessibility Guidelines (UAAG): Released in 2002, UAAG emphasized the role of browsers and media players in bridging the gap between accessible content and user needs.

These guidelines created a shared framework for developers, designers, and tool creators to prioritize accessibility. However, while transformative, they couldn’t fully account for the rapid evolution of web technologies and design practices.

Accessibility Is Not Yet the Default

Today’s accessibility challenges lie in bridging the gap between intent and implementation. Generative AI tools like GitHub Co-pilot and Cursor make integrating accessibility into development workflows easier, but they often require explicit prompts and lack proactivity. Testing tools like axe DevTools and Accessibility Insights simplify compliance checks but fall short of capturing nuanced user experiences. Meanwhile, overlays, often marketed as quick fixes, have sparked controversy for their superficial fixes and potential harm, as evidenced by the FTC’s recent fine against accessiBe. These challenges highlight the need for deeper engagement and thoughtful use of today’s tools to ensure inclusivity.

AI-Powered Development Tools: Progress with Limitations

Generative AI tools like GitHub Copilot and Cursor represent significant progress in embedding accessibility into development workflows. By generating accessible code snippets when prompted, they save developers time and effort. For example, they can generate ARIA-compliant labels or ensure proper keyboard navigation when asked to create “accessible components”.

However, these tools often require explicit instructions to prioritize accessibility, limiting their ability to proactively enforce best practices. Their success depends on the developer’s awareness and intent. Without proactive integration of accessibility principles, these tools risk perpetuating a reactive approach to inclusive design.

Testing Challenges: Going Beyond the Basics

Testing remains one of the most resource-intensive aspects of accessibility. Automated tools like axe DevTools from Deque Systems, Inc and Microsoft Accessibility Insights simplify the process by identifying technical issues such as missing alt text or color contrast problems. These tools provide actionable insights, helping developers address compliance issues earlier in the workflow.

However, these tools don’t fully capture the nuances of user experiences. Persona-based testing and real user engagement provide critical insights into how people with disabilities interact with digital content. The W3C’s guide on how people use the web emphasizes that understanding diverse needs requires direct collaboration and effort. Unfortunately, many organizations lack the resources to invest in this level of rigorous testing.

Overlays: Quick Fixes That Often Fall Short

Overlays have gained traction as tools that promise to fix accessibility barriers on-the-fly by applying adjustments to web pages in real time. While attractive for their simplicity, overlays often deliver superficial fixes that fail to address deeper design flaws. Worse, they can interfere with assistive technologies like screen readers and create usability issues for users.

The controversy surrounding overlays reached a tipping point recently when the Federal Trade Commission (FTC) fined accessiBe $1 million for misleading advertising. The company had claimed its overlays could guarantee full compliance, a promise that proved inaccurate and harmful. This incident underscores the dangers of relying too heavily on quick fixes instead of addressing root causes.

Accessibility as a Seamless Standard

The future of accessibility lies in leveraging generative AI and adaptive technologies to make inclusivity seamless and proactive. AI-powered editors could embed accessibility best practices by default, while user-agent-based overlays running on specialized models could offer dynamic, personalized adjustments for individuals. Agentive AI personas could scale testing by simulating diverse user interactions early in the design process. However, realizing this vision requires overcoming technical, ethical, and cultural barriers while ensuring technology complements, rather than replaces, the lived experiences of real users.

Generative AI Editors: Accessibility by Default

Imagine a future where accessibility isn’t just an option—it’s built into every stage of design and development. Generative AI editors, like enhanced versions of GitHub Copilot or Cursor, could proactively embed accessibility best practices into the codebase without needing explicit prompts.

Vision

Generative AI editors like GitHub Copilot and Cursor could evolve to embed accessibility best practices automatically. For example, they could apply ARIA roles, enforce keyboard navigability, and generate compliant patterns without requiring explicit prompts.

Benefits

Accessibility becomes the default, reducing reliance on developer expertise.

Eliminates common accessibility errors during coding.

Barriers to Implementation Today

Training data often lacks nuanced accessibility patterns.

Developers may resist ceding control to automated systems.

Potential Controversy

Some argue that automating accessibility could lead to complacency among developers, who might prioritize speed over understanding inclusive design principles.

User-Agent-Based Overlays: Cloud and Local Specialized Models

User-agent-based overlays powered by specialized language models (SLMs) offer a path to more personalized and effective accessibility solutions. Running locally on a user’s browser or device—or in the cloud for more complex computations—these overlays could dynamically adjust content to meet individual needs, such as resizing text, simplifying navigation, or enhancing contrast. This approach prioritizes privacy while tailoring solutions to users in real time.

Vision

Provide personalized accessibility features. These tools would adjust content dynamically—such as resizing text or optimizing navigation—based on individual needs.

Benefits

Real-time personalization.

Barrier detection and remediation.

Enhanced privacy for users if run locally.

Barriers to Implementation Today

High computational requirements for local models.

Lack of standardization across browsers.

Potential Controversy

Critics may view overlays, even adaptive ones, as masking deeper design issues rather than solving them.

Agentive AI Testing: Personas as Early-Stage Validators

Agentive AI personas represent an innovative approach to accessibility testing, simulating interactions from the perspective of users with diverse disabilities. These AI agents could emulate scenarios like navigating a site with a screen reader, interacting with only a keyboard, or processing complex layouts with cognitive constraints. This technology has the potential to catch barriers early in the design phase, reducing costs and improving usability.

Vision

Agentive AI personas could simulate interactions from the perspectives of users with various disabilities. For example, an AI persona might simulate a screen reader user navigating a complex form, flagging accessibility barriers early in the process.

Benefits

Scalable testing across multiple scenarios.

Early detection of barriers reduces rework costs.

Barriers to Implementation Today

AI personas cannot fully replicate lived user experiences.

Training such models requires extensive real-world data.

Potential Controversy

Relying solely on simulated testing risks overlooking nuanced insights from real users, reinforcing the importance of combining automation with human feedback.

Toward Accessibility by Default

True accessibility is achieved when inclusivity is embedded seamlessly into every stage of the design and development process. Generative AI tools, adaptive overlays, and agentive testing can accelerate progress, but they must be deployed thoughtfully. The future of accessibility depends on striking the right balance between automation and human insight, ensuring that real user experiences guide every step.

By leveraging technology responsibly and centering the needs of diverse users, we can create a digital world that is not only compliant but genuinely inclusive—for everyone, by default.