Figma adds real-time multimodal feedback and AI-assisted components

Design · 4 min read

The new multimodal feedback layer lets stakeholders drop voice comments or annotate screenshots directly in the canvas; Figma's server-side models transcribe, summarize, and link actionable items to frames. Designers reported that meeting friction decreased because decisions and context are preserved in a single place.

AI-assisted components can suggest responsive variants automatically when tokens or constraints change. For example, switching a primary color token prompts the component library to create high-contrast variants and update export assets. Figma says these components use a model to reason about accessibility, spacing, and interaction states.

Collaboration features include a 'design veterinarian' mode that flags inconsistent token usage, uncoupled instances, and accessibility issues in real time. Figma also rolled out an enterprise plan with centralized model governance and export controls for regulated teams.