Google Stitch Gets Massive AI Redesign With Gemini 3, 3D Workspace, and React Export

Google Labs has launched the biggest update ever to Stitch, its experimental AI-powered design tool, integrating Gemini 3 to deliver higher quality UI generation alongside a suite of new features that could reshape how designers and developers build interfaces.

Key Highlights

• Gemini 3 integration powers significantly improved UI generation quality
• 3D spatial workspace replaces the flat canvas with an immersive design environment
• Voice AI agent enables hands-free design through natural language commands
• Direct React export generates fully functional React applications from designs
• Advanced prototyping with screen-to-screen navigation and QR code testing

A New Kind of Design Tool

Stitch, which launched as a simple text-to-UI generator, is evolving into what Google calls a "vibe design partner." The tool now goes beyond static screen generation, allowing designers to stitch together different screens on a canvas to create working prototypes with real interactions and user flows.

The Gemini 3 upgrade brings noticeably better output quality. Designers can describe what they want in plain language and receive professional-grade interface designs in seconds.

3D Workspace and Voice Agent

Perhaps the most striking change is the shift from a flat canvas to a 3D spatial design environment. Users can visualize and manipulate designs with greater precision, which is particularly useful for complex UI/UX projects that require dynamic interactions.

A new AI agent appears as a compact, expandable widget that moves with users as they switch between design and prototype modes. The agent supports voice commands with multiple voice options and maintains an action log for transparency.

From Design to Code

The standout feature for developers is the ability to generate fully functional React applications directly from selected screens. This is not just a clickable prototype — it produces actual working code that can be exported to AI Studio or other development tools.

Google is also planning MCP export and agent skills integration, positioning Stitch within a broader AI ecosystem for seamless compatibility with other tools and platforms.

Impact on the Design Industry

The update has sparked significant discussion in the design community. Some see it as a direct challenge to tools like Figma, with one commentator noting that "when AI can design full UIs from a text prompt, the bottleneck moves from how it looks to what it should do."

For developers who are not designers, Stitch offers a way to produce professional interfaces without traditional design skills. For designers, it shifts the role toward systems thinking and experience architecture rather than pixel-level execution.

What's Next

Google notes that the feature remains experimental, with many improvements planned. The full redesigned version of Stitch is expected to be officially unveiled at Google I/O 2026, scheduled for May 19-20.

Stitch is available now at stitch.withgoogle.com with the Gemini 3 upgrade and prototyping features active.

Source: Google Blog