Figma: Make Kits and Make Attachments bring design systems into AI prototyping

What happened

On April 2, 2026, Figma released two updates to its Make feature: Make Kits and Make Attachments. Both address a persistent limitation of AI-generated prototypes — that they produce generic output disconnected from a team’s actual design system.

Make Kits allows design system owners to import npm packages, Figma library styles, variables, and tokens directly into Make. This means AI-generated prototypes in Make can now reflect the actual components, color tokens, and type styles an organization uses in production, rather than defaulting to placeholder styling. Teams can configure a kit once and share it across their organization, so every Make session starts with the same design system context.

Make Attachments lets users attach files — PRDs, brand guidelines, code snippets, images, and videos — directly to prompts within Make. Previously, teams had to re-paste contextual information each session. Attachments persist and give the AI richer input for generating work that fits a specific project or brief.

Context

The same release also included Built-in Approvals in Buzz (beta), available on Organization and Enterprise plans. This lets admins configure approval workflows for asset exports from Buzz, so design review sign-off can happen within the platform rather than through external communication.

These additions sit alongside Figma’s broader push to make AI generation inside its tools practically useful for organizations that already have established design systems, not just for quick-start prototyping.

Why it matters for designers

The gap between AI-generated prototypes and production-ready design has been a recurring friction point for design teams evaluating Make. Make Kits directly addresses this: by teaching Make how a team’s design system works, the AI has relevant constraints to work within, and the output requires less manual correction before it can be used in a real review cycle. For organizations with mature design systems, this makes Make a more viable option for early-stage exploration that connects to downstream implementation.