How GitHub's Licensing team used AI agents to champion accessibility

Published via GitHub Executive Insights | Authored by Bronte Van der Hoorn

At GitHub, accessibility isn’t just about compliance. It’s about ensuring every developer can do their best work. When the GitHub Licensing team needed to address critical accessibility improvements while maintaining their roadmap commitments, they turned to GitHub Copilot coding agent as an accessibility champion.

Prioritizing accessibility while maintaining momentum

The Licensing team faced a common engineering dilemma: important accessibility improvements from two audits needed addressing before a deadline. The work was substantial enough to potentially impact roadmap delivery, yet essential for creating an inclusive product experience.

Instead of the traditional approach of reallocating engineers, they asked: "What would this look like if agents were already part of the team?"

From concept to concrete results

The team delegated the outstanding accessibility issues to the Copilot coding agent in batches, with an engineer serving as reviewer and safety net across several short working sessions. The results demonstrated significant potential:

• 70% of issues: Smooth landings where the agent implemented correct solutions with minimal guidance
• 10% of issues: Needed human co-piloting to reach an acceptable solution
• 20% of issues: Required the engineer to take a more hands-on approach

Most importantly, the team resolved all accessibility issues without disrupting their roadmap commitments.

"Despite a learning curve and the assistance we needed to provide the agents, we achieved our aim of delivering both our roadmap and our accessibility goals."
Josh Schultz, GitHub Engineering Team Leader

Where agents excel in accessibility work

The experiment revealed clear strengths where AI agents can provide the most value to teams:

• Standardized improvements: Adding server names to links, fixing aria-label mismatches, and making heading structure changes
• Bounded modifications: Small, well-defined changes with clear success criteria
• Documentation and labeling: Ensuring proper annotations for assistive technologies

“This isn't about deprioritising accessibility. For us, it's about watching agents actually shoulder important work that makes our product more inclusive for everyone.”
Josh Schultz, GitHub Engineering Team Leader

How to integrate agents into your accessibility workflow

Our Licensing team's experience yielded several practical insights for teams looking to leverage AI agents effectively:

1. Batch for efficiency: Plan for some overhead with each agent interaction; batching reviews keeps things streamlined.
2. Use AI to prepare for AI: Copilot Chat can quickly transform raw notes into agent-ready tickets, streamlining the preparation process.
3. Fix small nits yourself: Tiny fixes are often faster to implement directly than through an agent feedback cycle.
4. Be precise in feedback: Crystal-clear pull request comments (e.g., "Remove Y first, then add X") increase the likelihood of agent success on the next attempt.
5. Proactive labeling: As you review your backlog, label agent delegatable items.

The multiplier effect

The GitHub Licensing team's experience demonstrates that AI agents can function as team multipliers, allowing engineers to maintain focus on high-priority roadmap items while still addressing critical accessibility work. This creates ways of working that aren’t simply additive but potentially multiplicative.

For engineering leaders, the implications can be transformative: AI agents aren't replacing accessibility expertise — they're increasing your ability to attend to this critical work without deprioritising your feature roadmap. By thoughtfully integrating agents into your workflow, particularly for well-defined accessibility tasks that might otherwise need to wait for the next sprint, you can maintain velocity on your roadmap while creating more inclusive products now.

In this new era of development, the teams that thrive won't just be those with the most resources, but those who most creatively harness the complementary strengths of humans and AI to deliver better, more accessible experiences for everyone.

---

Want to learn more about the strategic role of AI and other innovations at GitHub? Explore Executive Insights for more thought leadership on the future of technology and business.