Generative UI
A user interface where AI generates UI elements — components, layouts, even whole pages — in response to what the user is doing or asking.
In plain English
Generative UI is the idea that the interface itself can be dynamically composed by an AI model rather than pre-designed by a human. Instead of one chat box, the model produces the right widget — a form, a chart, a table, a card — for what the user is trying to do right now.
Two main flavours:
- Component selection — the model picks from a library of pre-built components and fills them with data
- From-scratch generation — the model writes HTML/JSX/Tailwind on the fly (powered by tools like Vercel v0, Lovable, Bolt)
Where you see it:
- ChatGPT Apps and Claude Artifacts — render interactive components inline
- Vercel v0, Lovable, Bolt — generate whole apps from a prompt
- Cursor / Windsurf — generate UI code in coding agents
- Block-based design tools that mix AI suggestions with editor controls
Open questions: Accessibility, performance, design consistency, and security all get harder when UI is generated per-session. Production deployments usually constrain generation to a vetted component set.