I recently wrote a piece for the zeroheight blog about something that's been on my mind: the growing risk that leadership sees AI generating UIs and concludes there's no need for a design system anymore.
The short version? They'd be wrong.
The gist
It's easy to conflate output with outcome. AI can generate interfaces fast, but without the guardrails of a well-structured design system, what you get is slop — things that look impressive in a demo but fall apart in production.
Design systems have never been about the code or the Figma files. They're the series of decisions people have painstakingly made through discussion, argument, and sometimes tears. Why a component works the way it does matters far more than the component itself, and that context is exactly what AI needs to generate anything useful.
What I cover
- Why design systems are more essential in the AI era, not less
- How tools like Model Context Protocol (MCP) servers for Storybook, zeroheight, and Figma give AI the context it needs
- A case study from a project where I used YAML-based component specs to generate front-end code — and how non-technical people could write those specs
- The emerging A2UI paradigm: applications signalling intent to AI, which generates interfaces from design systems at runtime
- Why the biggest criticism of design systems — that they slow teams down — might be about to disappear
The full article is up on zeroheight. Give it a read.
