Figma Ships 'Accessibility Insights' Plugin with On-Device AI for Color and Label Suggestions

AI · 4 min read

Accessibility Insights runs local models to analyze artboards and component libraries, then proposes color-pair swaps, alternative labels, and suggested ARIA roles. The plugin surfaces concrete changes — for example replacing a low-contrast button color or suggesting a descriptive label for an icon button — and shows why each change improves WCAG conformance.

A key selling point is on-device inference: Figma says the model executes in the browser or on the user's machine, reducing the need to send design files to cloud services. The plugin also integrates with token systems, so suggestions can be emitted as new or modified design tokens rather than one-off fixes, preserving system-level consistency.

Designers and accessibility leads appreciate the time savings, but some researchers caution that automated suggestions should be paired with real-user testing. Figma includes a feedback loop for designers to flag incorrect recommendations, and the plugin exports an audit report your team can attach to Storybook or the component repo.