Figma 2026.2 introduces real-time AI-driven component refactoring

Design · 5 min read

Figma's latest release embeds a real-time assistant that analyzes a project's component graph and proposes refactoring actions such as extracting variants, consolidating duplicates, and introducing responsive constraints. The assistant runs incrementally, surfacing suggestions non-destructively as design managers and engineers edit components.

The refactor suggestions come with rationale and change previews so teams can accept, tweak, or dismiss proposals. Figma is using a combination of supervised learning on anonymized component repositories and rule-based heuristics to avoid destructive automated changes and preserve intent.

Teams participating in early access reported a 20–40% reduction in duplicated components and faster onboarding for new designers. Figma also added a new audit view that highlights accessibility gaps and color-contrast fixes alongside the refactor recommendations.