Google's prompt-to-interface tool: describe a screen, get UI and front-end code to take further.
Stitch is Google's prompt-to-interface tool: describe a screen or paste a reference and it produces UI along with front-end code. Output can be iterated in conversation, exported to code, or pasted into Figma to finish by hand. It is built for the earliest part of the work, where the question is what the screen might be rather than what it should ship as.
- Prompt or reference to UI
- Front-end code included
- Iterate in conversation
- Paste into Figma to finish
- Free to use
- Mobile and web layouts
Recent X Posts
We're building a new kind of agent: a collaborative design partner that works alongside you. Jumping straight into coding agents often locks you into the first idea; starting in Stitch lets you rapidly explore variations to find the best idea. 🎨⚡ Watch this "Only Positive News" mobile app go from simple prompt to multiple UI iterations to a live AI Studio prototype in seconds. Turn the sound up! 🔊👇
DESIGN.md repo github.com/google-labs-co…
Four months ago, we launched the DESIGN.md standard to build a portable visual identity for agents. Recently, it hit one of the top trending repos on GitHub! 🤯 If you haven't checked the repo lately, it has leveled up. We just added: ✨ Linters ✨ Diff tools ✨ Translators Want to turn your DESIGN.md directly into a Tailwind theme? See the attached one line command. Keep watching this space. We have big plans for the future of OPEN and portable design formats. Repo link 👇
💡 Pro Tip: Need a pixel-perfect hero graphic? Ask Stitch to generate the image first as a standalone asset. This automatically routes your request to Nano Banana Pro, which is optimized for complex, professional use cases. Once it is on the canvas, just ask the agent to drop it into your screen.