Google Stitch major update adds infinite canvas, voice design agent and multi-screen generation

Google LLC today released a new version of Stitch, an artificial intelligence tool that can generate user interfaces for websites and mobile apps. The company said the tool uses its Gemini series of large language models to translate designs into HTML and CSS code while also supporting the Tailwind framework. The release adds a new interface centered on an AI native infinite canvas. This canvas can show multiple visual assets side by side to provide a complete view of a design project. Developers can instruct the tool to generate code for an interface section through a reference image or text description. The new version can produce up to five screens at once, whereas the prior release handled only one screen at a time. A play button allows users to preview how shoppers or other users will move through the generated screens in sequence. Josh Woodward, vice president of Google Labs, explained in a blog post that Stitch can automatically generate logical next screens based on the click. The updated tool supports voice commands for refining design elements including font size. A newly added AI agent powers several of these capabilities. Stitch can also export interface elements to external tools and save design details in a natural language file called DESIGN.md. The tool is available for free through Google Labs.