Google has unveiled 5 major new upgrades to its impressive design tool Stitch:

1. AI-Native Canvas - a completely redesigned, node-based infinite canvas that can hold and reason across images, code, and text at the same time. You can drop in images, PRDs, or code as creative starting points. There's also a new Agent Manager to run multiple design ideas in parallel, plus a long-requested light mode.

2. Smarter Design Agent - the AI now understands your entire canvas context, not just isolated prompts. You can ask it to swap logos, generate product briefs from existing screens, interview you to define a new design, or create mobile versions of desktop screens.

3. Voice Design ("Vibe Design") - in preview, you can talk to Stitch while it literally sees your canvas and selected screens. Use it to request critiques, ask for variations, or navigate around.

4. Instant Prototypes - hit play and get a working prototype in seconds. Stitch can even predict and generate the next screen based on where you click, making interactive flows much faster to build.

5. design.md & Design Systems - every new design automatically gets a cohesive design system applied for consistency. The new design.md file acts as a portable design rules document - you can export your design tokens/rules and import them into other projects, similar to how developers use config files.

Impressive stuff from the product teams at Google Labs - and yet another example of a product adding Voice features that reflect the way UX is changing in 2026.

More here: x.com/stitchbygoogle/st…