Developers

Your workspace, inside your AI assistant

Point Claude, ChatGPT or your editor at the MCP server and ask what visitors wanted, what went unanswered, and who's waiting for a person.

The endpointStreamable HTTP · Bearer token
https://api.integrable.cloud/mcp

The tools

Nine tools. Eight read, one writes.

Read

get_analytics

Conversations, leads and outcomes for a period.

Read

get_usage

Credits used and remaining.

Read

list_assistants

Every assistant in the workspace.

Read

get_assistant

One assistant's settings.

Read

search_conversations

Find conversations by what was said.

Read

get_transcript

A whole conversation, turn by turn.

Read

list_handoffs

Visitors waiting for a person.

Read

list_knowledge

The sources an assistant answers from.

Write

add_knowledge

Add a document to close a gap.

Connecting

Three steps, about a minute

  1. 1

    Create an API key

    Scoped to what the client should be able to do.

  2. 2

    Add the server

    Your client takes the URL and the key.

  3. 3

    Ask it something

    “What couldn't my assistant answer this month?”

  • Which clients can connect?

    Any MCP client that can connect to a remote server with a token — Claude, ChatGPT and MCP-capable editors. Per-client setup is in the docs.
  • Can it change my configuration?

    No. Eight of the nine tools only read. The one write is add_knowledge, which adds a document to an assistant's knowledge base — and it counts against your plan like any other document.
  • Where does my data go?

    Whatever the client reads goes to the model that client uses, which may not be one of ours. Connecting a client is you choosing that model provider.

Free plan · no card

Get a key and connect it

Free plan, no card. Connecting takes about a minute once you have a key.

  • Live in minutes
  • Answers cite your pages
  • Cancel any time