Google Stitch 2.0: Your Vibe Design Partner With New AI Agent, Infinite Canvas, and DESIGN.md

The way we design software has changed fast since we started incorporating and using artificial intelligence (AI) into production. A task that once required weeks of wireframing, feedback loops between designers and developers, and dedicated prototyping tools is now crammed into a natural-language conversation with an AI canvas. Google is pushing this change forward with a major update to Stitch, its AI-native UI design platform under Google Labs.

Stitch is now your vibe design partner, an AI-native software design canvas that allows anyone to create, iterate, and collaborate to turn natural language into high-fidelity UI designs. This update changes what a design tool is actually supposed to do for designers, developers, and founders.

What is Stitch by Google?

Stitch was introduced at Google I/O 2025, following Google's acquisition of Galileo AI and its rebranding. The original version was relatively straightforward; you type a prompt, receive some screens, export to Figma, or grab the generated HTML and CSS. Useful for the first 80% of ideation, but not much else. The March 2026 update is much bigger than baseline, introducing five major additions at once.

1. The AI-Native Infinite Canvas

The structural centerpiece of this update is a completely redesigned workspace. The new interface features an AI-native infinite canvas that gives ideas room to grow from early ideations to working prototypes, and can accept different types of input (text, images, and code). This is meaningful because the AI works with all of those inputs simultaneously, and you are not limited to text prompts alone.

Instead of starting with a wireframe, you can start by explaining the business objective you are hoping to achieve, what you want your users to feel, or even examples of what is currently inspiring you.

The canvas's breadth of context feeds directly into a redesigned design agent:

• Smarter design agent: A new design agent works alongside the canvas and can reason across the entire project's evolution.
• Agent Manager: A new Agent Manager can track progress and help users stay organized while working on multiple ideas in parallel.
• Parallel exploration: Run multiple design directions at the same time without losing earlier work or progress.

2. Voice Design, Powered by Gemini Live

One of the most unique new additions is voice input, currently available in preview. In addition to typing, users can use voice commands to talk to the canvas, which is powered by Gemini Live. The agent can see the canvas and selected screens in real time.

The agent can give real-time design critiques, design a new landing page by interviewing you, and make real-time updates like "give me three different menu options" or "show me this screen in different color palettes" as you speak. The intent is less about replacing keyboard input and more about keeping designers in a creative flow without having to break to rewrite a prompt from scratch.

3. Instant Interactive Prototypes

Static screens have always been a limited proxy for real user experience. Stitch can address this directly with one-click prototyping:

• Users can click the "Play" button to preview the app flow, and Stitch can automatically generate the next logical screens based on a click to map out the user journey.
• Stitch can convert static designs into clickable prototypes in one step, connect screens, hit Play, and walk the user journey.
• Designs can be exported directly to Figma with editable layers and Auto Layout, or downloaded as HTML, CSS, and React code.

4. The Smarter Design Agent

The new design agent is the intelligence layer powering everything on the canvas, and it has been substantially rebuilt for this update. The design agent has been rebuilt to understand the full canvas context, not just the currently selected frame. This is a meaningful distinction; earlier AI design tools processed one screen at a time, losing sight of earlier decisions as the project grew.

What makes the agent genuinely different is its interview mode. Rather than asking you to describe what you want, the agent asks questions to surface your actual business objectives, how you want users to feel, and what is inspiring you. A new design agent tracks the entire history of a project, allowing it to reason through different versions. That project-wide memory is what allows consistency at scale, a problem that has historically made AI-generated UI feel fragmented and difficult to build upon.

5. DESIGN.md — The Feature That Connects Design to Development

The most strategically significant addition in this update may also be the least visually dramatic. DESIGN.md is a markdown file that captures design rules (colors, typography, spacing, component patterns) in a format that humans and AI agents can read.

Users can extract a design system from any URL and use the DESIGN.md agent-friendly markdown file to export or import design rules to or from other design and coding tools. This ensures there is no need to rebuild the same design tokens in every new project or tool. Using the recently released Stitch MCP server and SDK, designs can be exported to developer tools such as AI Studio and Antigravity, ensuring a seamless partnership among the user, the AI, and developers.

Availability and Pricing

Stitch is currently completely free, with no paid plans announced. The tool is available at stitch.withgoogle.com for users 18 and older in every region where Gemini is supported, and no credit card is required, only a Google account.  

• You can use Gemini 3 Flash to generate high-quality designs in HTML and export to coding agents.
• There is also an option to use Gemini 3.1 Pro, which prioritizes maximum quality and reasoning over speed.

The competitive dynamics here are less about which tool makes prettier mockups and more about who controls the pipeline from idea to deployment. While Figma does own the professional design workflow, Google is building toward an AI-native one where the starting point is language, not layers.

Design a dark-mode SaaS analytics dashboard for a B2B AI tool. Include a sidebar navigation, KPI cards showing monthly active users, revenue, and churn rate, and a line chart. The tone should feel modern and data-dense, like Linear or Vercel.