Google Stitch: What it does, how it works and other details about the new AI-powered tool

Google Stitch

At Google I/O 2025, the company unveiled Stitch — an AI-powered design assistant that lets developers generate frontend code and user interfaces with just a few words or an image. It’s Google’s latest play in the rapidly expanding world of AI-assisted, or “vibe,” coding.

According to Google, Stitch is an AI-powered tool to help application builders generate high-quality user interfaces for mobile and web apps, and easily export them to Figma, or directly access the frontend code. It uses some of the latest AI models from Google DeepMind.

Stitch is available for free of charge with certain usage limits. Each user receives a monthly allowance of 350 generations using Flash mode and 50 generations using Experimental mode. “Please note that these limits are subject to change,” says Google.

Stitch is designed to streamline the early stages of app and web design. Prompt it with a phrase like “photo-sharing app” or upload a rough sketch, and Stitch will spit out HTML and CSS code to match. Users can pick between Google’s Gemini 2.5 Pro or the faster, more lightweight Gemini 2.5 Flash model to power their workflows.

While not as expansive as some of its rivals — like OpenAI’s Codex or GitHub Copilot, which just got a fresh batch of updates at Microsoft Build — Stitch isn’t trying to replace a full development stack. Instead, it focuses on UI generation, offering smart design layouts with enough flexibility to tweak and refine. The tool supports direct export to Figma and allows devs to open the code in their IDEs for further editing.

Google says that it is working to bring its tools to as many people as possible. “The site is currently available in English only to 18+ users in 212 countries,” says Google.