AI design and prototyping tools compared
By Imran Sheikh · 2026-11-12 · Models and tools · 4 min read
Key points
- Canvas tools explore; app builders ship working prototypes.
- Design rules travel between tools through files like DESIGN.md.
- Use one tool per stage, and hand off to a coding agent.
Two years ago an AI design tool meant a plugin that renamed your layers. Now you can describe a product and get a clickable prototype, a working app or a set of on-brand screens in minutes. Making something is quick now. Choosing which tool to open for which stage of the work takes more thought.
This post compares the AI design and prototyping tools we see most, as of November 2026. We group them by the job they do best, then show how they fit together in a real workflow.
Two kinds of tool
The tools split into two groups. Design canvases help you explore and refine how something looks and flows: Figma with its AI features, Google Stitch and Claude Design. App builders turn a prompt into running code you can deploy: v0, Lovable and Bolt. Claude artifacts sit in between, as small working tools you make inside a chat. The lines blur, since every canvas now exports code and every builder has a visual editor. But the starting point still shapes the result.
| Tool | Best at | Output | Entry pricing |
|---|---|---|---|
| Figma (Make, agent, MCP) | Working inside an existing design system | Designs, prototypes, web apps | AI credits per seat; 3,000 a month on a Professional full seat |
| Google Stitch | Fast UI exploration on an infinite canvas | Screens, prototypes, code, DESIGN.md | Pricing not stated in Google’s announcements |
| Claude Design | Prototypes, decks and one-pagers by conversation | HTML, PDF, PPTX, Canva, Claude Code handoff | Research preview on paid Claude plans |
| Claude artifacts | Small interactive tools and shareable pages | Hosted pages that can call Claude | Free and paid Claude plans |
| v0 | Full-stack web apps on Next.js and shadcn/ui | Code deployed to Vercel | Free with 7 messages a day; Plus $30 per user a month |
| Lovable | Apps with a built-in backend | Hosted apps | Free daily build credits; paid plans add monthly credits |
| Bolt | Apps with hosting and a database included | Hosted apps | Free with 1M tokens a month; Pro $25 a month |
Design canvases
Figma now ships AI across the product. Figma Make turns prompts and existing Figma designs into prototypes and web apps. A Figma agent can generate and remix designs and give design feedback. The MCP server lets coding agents read your designs directly. All of it draws on a monthly credit pool: 3,000 credits for a full seat on Professional, and 500 for Collab, Dev and View seats. If your design system already lives in Figma, this is where AI output will match it most closely.
Google Stitch was rebuilt in March 2026 as what Google calls an AI-native software design canvas. It has an infinite canvas, a design agent that can run several ideas in parallel, and interactive prototypes you can play through. It also introduced DESIGN.md, a markdown file that holds design rules in a form agents can read. In May, Google added real-time design that streams to the canvas, plus exports to Google AI Studio, Antigravity and Netlify. Stitch is quick for broad exploration, when you want many directions before you commit.
Claude Design is Anthropic’s design product, launched on April 17, 2026, as a research preview for Pro, Max, Team and Enterprise subscribers. You describe what you want and refine it in conversation. It exports to Canva, PDF, PPTX or standalone HTML. When a design is ready to build, it packages a handoff bundle for Claude Code. That handoff is the part we value most, because the agent gets the design intent along with the screens.
App builders
v0 from Vercel describes itself as an AI agent that helps anyone create real code and full-stack apps. It works with Next.js, Tailwind CSS and shadcn/ui, and deploys to Vercel. That makes it a strong pick when the prototype might become the real web product. The free plan allows 7 messages a day. Plus is $30 per user a month and includes $30 of monthly credits.
Lovable and Bolt aim at the same person: someone who wants a working app with login and data, without setting up a backend. Lovable gives free users 5 build credits a day, up to 30 a month, plus a monthly grant of Cloud credits for hosting and data. Bolt’s free plan includes 1M tokens a month, and Pro at $25 a month starts at 10M tokens with hosting and custom domains. Bolt can also start from a Figma file or a GitHub repo.
Claude artifacts fill a smaller gap. They are pages and tools you make inside a Claude chat, like a pricing calculator for a pitch or a quick survey. Artifacts can call Claude directly, and the people using them sign in with their own Claude accounts, so you don’t manage an API key. They start private and can be shared when you choose.
Where each fits in a real workflow
Here is how a project tends to move through these tools in our studio. Projects bend it, but the order holds most of the time.
- Explore: Stitch or Claude Design for many quick directions. Low cost, fast to throw away.
- Decide: bring the chosen direction into Figma, where the design system and the team’s comments live.
- Test: a clickable prototype from Figma Make or Stitch for usability sessions. If the test needs real data, a quick Lovable or Bolt build.
- Build: hand the design to a coding agent like Claude Code through Figma’s MCP server, a DESIGN.md file or Claude Design’s handoff bundle.
- Share: a Claude artifact or a v0 deploy for stakeholder demos and small internal tools.
How we choose
We ask three questions. Where does the design system live? That usually means Figma, and anything that reads Figma well. Will this prototype become the product? If yes, we start in code with v0 or a coding agent, so nothing is thrown away. Who needs to touch it next? A designer needs a canvas, an engineer needs a repo, and a client needs a link. Pick the tool that hands off cleanly to that person. The tools will keep changing. Clean handoffs between stages are what keep a project moving.