Google Stitch

Overview

Google Stitch is an experiment from Google Labs that turns prompt and image inputs into UI designs and the frontend code behind them. It launched in May 2025 on the multimodal capabilities of Gemini 2.5 Pro and now runs on Gemini 3, which Google describes as an AI-native software design canvas: it takes natural language, images, text, code and voice, returns high-fidelity screens, offers multiple design variants to compare, and refines them through an interactive chat and theme selectors. A Prototypes feature links individual screens into interactions and whole user flows rather than leaving them as static mockups, with logical next screens generated automatically. Designs leave the canvas three ways: as generated frontend code, through a paste-to-Figma function, and through a portable DESIGN.md format that carries a design system between tools. For programmatic use there is a Stitch API behind an official Node.js SDK and an MCP server, authenticated by API key or OAuth, which generate screens, edit them, create variants and extract HTML and screenshots.

Best for product and engineering teams that want a first high-fidelity screen and its frontend code from a prompt, then hand it on to Figma or a coding agent.

Stitch sits at the point where a design conversation becomes something a developer can open, and it is unusually well suited to being driven by another tool rather than only by a person: the MCP server and SDK mean a coding agent can commission a screen the same way a designer would. That makes it interesting less as a replacement for a design tool than as the first step in front of one. The caveat is its status. It is a Labs experiment, the SDK repository states plainly that it is not an officially supported Google product, and Google publishes no plan structure or usage limits in a form that can be read outside the app, so anything built on it should assume the ground may move.

Key Facts

Features

What it makes

High-fidelity UI screens and interactive prototypes that link screens into user flows, plus the frontend code behind them

AI generation

Generates and edits UI from natural language, images, text, code and voice input, on Gemini 3 and on Gemini 2.5 Pro at launch, with multiple design variants to compare and voice-driven design critiques

Brand, style and templates

Theme selectors, and a portable DESIGN.md format that carries a design system between tools

Export and handoff

Frontend code generation, a paste-to-Figma function, handoff to Google AI Studio and Antigravity, and HTML and screenshot extraction through the SDK

API

Yes. The Stitch API is reached through an official Node.js SDK and an MCP server, authenticated by an API key or OAuth, and generates screens, edits them, creates variants and extracts HTML and screenshots. The SDK repository notes it is not an officially supported Google product

Integrations

• Figma
• Google AI Studio
• Antigravity
• Gemini CLI
• Claude Code
• Cursor

Deployment and Support

Deployment

• Cloud (SaaS)
• Browser-based

Onboarding

• Documentation and knowledge base

Pros and Cons

Pricing

Best For

• Agencies
• Media & Creative
• Technology

Alternatives and Comparisons

What users say

Reviews

Share your experience

Write a Review