Introduction
Figma has spent the past year turning into more than a canvas. With Figma Make and its AI agent, teams can now generate layouts, populate components, and prototype directly inside the file. That's a real shift in how design work gets made.
But an AI agent working inside Figma still only sees what Figma sees: components, variables, tokens, and whatever documentation happens to be open. It doesn't know why the brand looks the way it does, or what it's trying to say.
What this tool does well
Figma remains the strongest environment for design systems. Components, variants, and variables give teams a shared, structured library of what the brand looks like. Figma Make extends that with AI-assisted generation, letting teams move from component to layout faster than before. For visual consistency at the asset level, Figma is well built for the job.
Where Ai still lacks context
Figma is excellent at product design systems. A button has defined states, a spacing token has a defined value, an onboarding screen either follows the pattern or it doesn't. That's a largely linear problem, and Figma's agent handles it well.
A brand system isn't linear in the same way. It's look, feel, and communication: why a color is supposed to read as calm rather than energetic, what tone a sentence should carry next to that button, what an image is meant to make someone feel. Figma's agent can reach into a file and pull the right token. It can't explain why that token is the right one, or what story it's supposed to be telling.
This isn't a Figma-specific gap. It shows up across the industry. The State of AI Design 2026 report found that 42% of designers cite lack of product or brand context as a top challenge when using AI tools for design work, ranking above integration difficulty, security concerns, and cost. Some enterprise teams have started building internal tools to close this gap themselves. Figma's component library solves the product design system problem. It was never built to solve the brand system problem, and it isn't a criticism of Figma to say so.
Why brand context matters
Brand Context is structured across three layers, and each one answers a different question Figma's agent can't answer on its own.
Precision gives the agent exact values: hex codes, spacing scales, font families, file references. This overlaps with what Figma variables already store, and keeps the two systems aligned rather than duplicated.
Semantic gives the agent the reasoning behind those values: why this green reads as "deep, muted, cool-toned" rather than a generic forest green, why this typeface pairs with that one, what mood a layout should carry. This is the layer a component library has no way to express.
Relationship gives the agent the rules that govern how everything fits together: which logo variant to use against which background, what a color's approved and forbidden pairings are, which token a background color must be paired with for accessible contrast. These are the decisions a brand team already made once. Brand Context means Figma's agent doesn't have to guess them again on every file.
It helps to think of it this way: logo, color, type, and motion aren't the brand. They're the actors, the sets, and the lighting that carry a story someone already decided to tell. Figma's component library can describe the actors in precise detail. It has no way to describe the story they're in. That's what the Semantic and Relationship layers exist to carry, and it's why this is a different kind of problem than keeping a design system tidy.
Sameness delivers this through an MCP server and a brand.json manifest, so the same brand definition that publishes as a website or PDF is also queryable by any AI tool sitting inside a workflow like Figma's.
Example workflow
Brand Context (Sameness) ↓ Figma agent queries Precision + Semantic + Relationship layers ↓ Agent selects the correct component and token combination ↓ Agent generates the layout using Figma's design system ↓ Output is checked against brand principles (color pairing, logo clearspace, type hierarchy) ↓ Consistent, on-brand result — without a designer re-typing brand context into the prompt every session