Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.summand.com/llms.txt

Use this file to discover all available pages before exploring further.

​
Prerequisites

  • A Summand account at summand.com
  • An OAuth-aware MCP client — claude.ai, ChatGPT, or any other client that supports remote MCP servers

​
1. Upload a dataset

  1. Sign in at summand.com.
  2. Create a connector — upload a CSV or connect a database.
  3. Wait for the analysis pipeline to finish. The dataset shows as Ready when its semantic layer is built.
Each dataset gets an ID of the form ds_019… (UUIDv7). You’ll use these IDs when referencing datasets in your assistant.

​
2. Connect your AI client

Point your client at the Summand MCP endpoint: 
https://mcp.summand.com/mcp
Most clients only ask for the URL — they handle the OAuth sign-in for you. See the connect a client page for step-by-step claude.ai and ChatGPT setup.

​
3. Try it out

Once connected, ask your assistant:
List my Summand datasets.
The client will call the list_datasets tool and return your datasets with their IDs, status, and last-updated time. From there you can drill into columns, surprise findings, EBM graphs, and more.
See the tools reference for every prompt-friendly capability the server exposes.

​
Going further

If you want to call the Summand REST API directly (for scripts, automation, or your own integrations), generate an API key from your account settings.