Skip to content

Microsoft BuildJun 3, 2026

Microsoft Build accessibility scanner booth demo

More than 1,000 people visited the Microsoft Accessibility booth, with 700+ diving into demos and conversations around accessibility and AI.

demoAccessibilityAI/MLAT testingCI/CDDevEx
1 / 4

GitHub + Microsoft Accessibility booth at Build

At Microsoft Build, GitHub and Microsoft created a hands-on Accessibility booth where attendees explored how AI can help identify, prioritize, and remediate accessibility issues. Across four stations, attendees selected accessibility personas, reviewed scanner results, and used Copilot Coding Agent to work through remediation workflows in a demo repository.

The experience generated 5,334 Copilot requests, consumed more than 204 million tokens, and triggered 228 Copilot coding-agent jobs. Those jobs represented approximately 93% of GitHub's coding-agent activity across the expo floor (228 of 246 total). The booth also drove more than 88,000 engagement events. In the demo repository, attendees created 2,520 issues and used 9,659 GitHub Actions minutes across 8,470 jobs.

Estimated attendance was approximately 650 badge scans across two days—the highest in the category—with the Build pin gamification track contributing to traffic and awareness. Attendees used the booth in two distinct ways: some completed a quick, hands-on interaction, while others stayed 30–45 minutes or longer to explore multiple personas, review scanner results, and discuss accessibility and AI-assisted remediation.

Accessibility scenarios

The persona-based flow invited attendees to consider a range of accessibility experiences:

  • Jordan: low vision (908 selections)
  • Alex: colorblind (617 selections)
  • Maya: blind screen reader user, navigation-focused (465 selections)
  • Sam: keyboard-only (308 selections)
  • Devon: deaf (135 selections)
  • Sasha: blind screen reader user, image and graph alternatives (84 selections)

The distribution offered useful signals across low vision, color blindness, screen reader usage, keyboard-only navigation, deaf and hard-of-hearing experiences, and image-description workflows. Jordan, Alex, and Maya appeared first in the interface, which likely influenced selection patterns; Sam's flow was also unavailable for approximately two hours on day two before the issue was resolved.

The project showed how AI-assisted accessibility work can create meaningful conversations about inclusive design and accessibility engineering while helping teams move from awareness to action.

Explore the live demo Microsoft Build booth accessibility scanner demo (opens in new tab).