Blueprint Agent
Describe a project, and an AI agent builds it with you in an in-browser IDE. Then you ship it. The Blueprint Agent (at ai.tangle.tools) is Tangle’s AI coding product: the build half of build, run, and improve.
What it is
An AI programming workspace in the browser. You describe what you want, the agent writes and edits the code, and you review, run, and iterate on the result in a real dev environment, with no local setup. Each session runs on the Sandbox runtime, so the agent has a full machine: shell, files, ports, and the coding agent runtime.
It does not improve itself. You drive every change; the agent does the work you direct. (The trace-and-improve loop lives in Intelligence, a separate product.)
Why use it
- Go from idea to running project without a local setup. Describe it in plain language; get working, reviewable code in a live environment.
- Keep the whole build in one place. Prompt, code, diffs, and a running preview sit in the same session, so you can see what the agent did and correct it.
- Build against real briefs. Partners post what they want built; the workspace surfaces that work for developers and agents to pick up and deliver.
How it works
- Describe the project: a plain-language brief of what you want built.
- The agent builds it: it works in a sandboxed dev environment, writing and running code, with the model you choose.
- Review and iterate: read the diffs, run the preview, and steer the next turn.
- Ship: take the result to your own repo or deployment.
Who it’s for
- Developers who want an agent to draft and iterate on a project in a real environment, not a chat window.
- Partners with concrete build briefs they want delivered by developers and agents.
Start
- Quickstart: start a session and get your first project building.
- Profiles: set model routing, tool permissions, and MCP servers for a session.