Nimble AI launches a code-to-design model that generates UI components from TypeScript props

AI · 5 min read

The Nimble model accepts component definitions, prop types, and optional storybook knobs, then produces a visual mockup along with HTML/CSS (or framework-specific JSX) and accessibility attributes. It learns from mapping patterns between prop semantics and common UI affordances, reducing back-and-forth between engineers and designers.

Team leads report that the model speeds internal demos and helps align expectations early in the sprint. Nimble includes guardrails to avoid inventing unsupported props and provides a confidence score for generated UI to help developers decide when to iterate manually.

Nimble offers an on-prem runtime for teams with IP concerns and a hosted version with integrated preview hosting. The company hopes to broaden language support beyond TypeScript and add design-system-aware generation so output conforms to an organization's tokens.