Developers

Build with the world model.

Three surfaces over the same graph: an MCP server for your agents, a sandbox tenant for your tests, and a read API for your code. All cite back to source.

MCPPreview

Bring Channl into your agents.

An MCP server that exposes the world model as a tool. Your agents query accounts, signals, and entities the same way a customer does, with citations attached to every answer.

  • Search accounts and signals, resolve entities, pull citation chains
  • Scoped tokens with per-tenant rate limits and audit logs
  • Works with Claude Desktop, Cursor, and any MCP-compatible client
Request MCP access
SandboxComing soon

Try the graph on a fixture set.

A sample tenant with seeded accounts, signals, and entities. Test your queries, prompts, and integrations against a deterministic dataset before pointing at production.

  • Deterministic seed data covering KYC, GTM, cyber, and intelligence
  • Reset endpoint that returns the tenant to a known-good state
  • Same auth, schema, and rate limits as production
Get on the sandbox list
APIPreview

Read the world model directly.

Authenticated, rate-limited endpoints over accounts, signals, citations, and the resolved entity graph. Every response cites its primary source.

  • REST + JSON. Standard pagination and caching. Idempotent writes
  • Per-key scopes, per-tenant audit log, signed webhook events
  • OpenAPI spec and Postman collection on request
Request API access

Need a different shape?

If you are integrating Channl into a product, a research workflow, or a workflow we have not thought of, write a short note.barry@channl.ai

← Back to home