What is Atelier
Atelier is a production-grade platform for AI builders. Three surfaces on one project.
- Studio — the visual builder. Plan, design, generate, test. Idea → schema → code → preview.
- Base — the backend runtime. Database, auth, storage, functions, realtime, LLM. Multi-tenant from minute one.
- Tooling — programmatic access. CLI, MCP server, integrations for Cursor, Claude Code, Codex, OpenCode, Windsurf.
The three are parallel ways into the same project. Studio is first-class but not the only client; the project’s state lives in Base, and any client — Studio, CLI, MCP, your editor — reaches the same surface.
Start here
- Quickstart — idea to deployed app in five minutes, walking through Studio, Base, and a deploy.
- Production from minute 1 — the inventory of what’s wired the moment you ship: auth, canary rollouts, observability, billing, more.
Why Atelier
Other AI builders ship demos. The hardest week of building modern software isn’t the prototype — it’s the glue between a working prototype and something you can charge for. Domain, auth, multi-tenant data isolation, deploys that don’t break users, error tracking, analytics, billing.
Atelier ships SaaS. Every new project starts on the other side of that week. Studio writes code against a runtime that already knows how to be production. From idea to first paying customer, no tools between.
Pick a surface
| Surface | Use it when |
|---|---|
| Studio → | You want a visual builder for planning, authoring, design system bindings, and tests. Idea-to-app with AI in the loop. |
| Base → | You want to understand the backend SDK your app runs against. Database, RLS, auth, functions, LLM, storage, realtime. |
| Tooling → | You want to reach Atelier from outside Studio — terminal scripts, AI agents over MCP, or external editors like Cursor. |