Figma announces new features to define how AI agents work on its platform, including Figma Specs, a Markdown editor that lets users define their design language

Your design system documentation can now directly guide AI agents, reducing manual oversight and ensuring brand consistency in AI-generated outputs.

Figma has introduced new features aimed at defining how AI agents operate within its design platform. A key announcement is Figma Specs, a Markdown editor that allows users to document their company's design language, serving as a reference for both human teams and AI agents. This feature can be a standalone file or integrated within the design canvas, supporting the embedding of design elements, FigJam diagrams, code layers, and existing designs. Figma also aims to integrate these specifications with its design and code libraries, and the Payload headless CMS. The company also expanded the public beta of its code layers feature, enabling users to add interactivity to prototypes and convert designs into code. Figma's Chief Design Officer, Loredana Crisan, highlighted the platform's evolving role in software development and design, noting the increase in code production versus product releases. The company also announced support for over 20 new tools and integrations.

Figma's new features, particularly Figma Specs, signal a maturing approach to integrating AI into creative workflows. By providing a structured, machine-readable format for design language, Figma is attempting to mitigate the unpredictable nature of AI and steer it toward predictable, brand-aligned outcomes. This move is critical for enterprises relying on consistent branding and design systems across a multitude of outputs, especially as AI-generated content scales.

The integration of Figma Specs with code and design libraries, alongside the expanded public beta of code layers, suggests Figma is positioning itself as a more comprehensive command center for product development. This enables a tighter loop between design, prototyping, and initial code generation, potentially streamlining the handoff process and allowing for more rapid iteration based on AI-assisted feedback. The support for external AI tools further indicates a strategy to become an ecosystem hub rather than a closed-off design application.