Who Owns the Context?

Figma’s recent examples of teams working with its agent are worth paying attention to.

• At Uber, agents help document components and map colors to design tokens.
• At Atlassian, motion becomes a reusable capability rather than a specialist handoff.

These aren’t just examples of AI making designers faster. They show what happens when a design system becomes something an agent can actively work with. And they raise an interesting question about Figma’s evolving role.

For years, the design tool was where we created interfaces. Now, AI can generate interfaces, write code, and build working experiences without ever opening a design tool. Figma’s response is to bring agents into the canvas, giving them access to the systems, libraries, and guidelines that make a team’s work distinctive.

Figma has a lot of that context. But there’s another step.

A design system has traditionally been a set of guidelines and components that people interpret and apply. Now agents can read those guidelines, assemble components, and generate new experiences. Great, the agents have context. However next is making that context authoritative, keeping it current, and validating the output. And the system doesn’t stop at the design canvas. It extends into code, content, and the experiences people use.

As platforms race to own the relationship between agents and the context they need, we think there’s an important distinction: the system itself shouldn’t be locked inside any one tool.