Stitch 3.0 by Google - Generate and iterate UI screens with AI on a live canvas
by•
Stitch generates UI screens for mobile and web from text prompts, with streaming edits, in-place AI changes, and one-click export to Figma, Netlify, Lovable, and Bolt.
For product designers and developers prototyping fast.
Replies
我只需描述 “一款极简记账 APP，主色调为薄荷绿，强调轻松、无压力的理财体验”，Stitch 就能生成符合整体氛围的完整界面，包括色彩搭配、字体选择和交互逻辑。上周我用它为客户快速生成 3 种不同风格的原型（简约风、活泼风、专业风），客户当场选定了方案，节省了至少 3 轮沟通时间
Generating UI screens on a live canvas is where this gets interesting for mobile devs most AI design tools are still web-first. Does it export to SwiftUI or Xcode-compatible formats or mostly stays in the design layer?
So far the expirience with generating some samples for my personal WebApp project were nice in Stitch 2.0, am excited to try and test out 3.0 as well, would try the same prompts and see how much it has improved since it took a few tries before even though mentioned in prompt for the design style yet it seemed to have missed a bit before some elements
What stands out to me is the shift from “AI generates screens” to “AI understands existing product context.” The codebase/Figma sync + in-place editing workflow feels far more useful than one-shot generation tools. Curious to see how reliable the component consistency becomes at scale.