Build and run AI agents on Tangle

Tangle provides five layers for AI agents: Sandbox, Runtime, Router, Integrations, and observability through Intelligence. Evaluation is part of each layer and the products built on them.

Start with a live service

  • Sandbox

    Run an agent in an isolated computer with files, a shell, and a browser.

    Live service
  • Router

    Route model requests through one API.

    Live service
  • Intelligence

    Send agent traces to inspect runs and compare changes.

    Live service

Hosted assistants are a preview

Tangle's first packaged use case is a hospitality assistant. Agent Builder is expected to provide the creation and approved channel setup path. The first completed text reply and full request-to-resolution flow have not yet been verified.

See how the five layers fit in the infrastructure architecture. For an in-browser coding workspace, see the Blueprint Agent. To inspect agent runs and compare changes, see Intelligence.