Overview
- The update adds an infinite canvas, a project‑aware design agent and an Agent Manager, with the system powered by Google’s Gemini multimodal models.
- Users can generate high‑fidelity, interactive app screens and front‑end code directly from natural‑language input or spoken requests.
- New voice features enable real‑time critiques, interviews to shape layouts, and live edits such as alternate menus or color palettes while you speak.
- Interoperability now includes DESIGN.md for portable design rules, a Stitch MCP server and SDK for skills and tool integrations, and exports to Figma, AI Studio and Antigravity.
- Stitch remains a Google Labs experiment with free daily credit limits, English‑only access for adults in Gemini‑supported regions, and Figma’s stock fell about 8% after the announcement.