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.

Ask Summand is a chat interface for your datasets. Type a question in plain English and Summand answers using the dataset’s semantic layer — running SQL, building charts, and citing the rows behind every claim.

​
What you can ask

  • Exploratory questions. “Which features matter most for predicting churn?” “What’s the median order value by region?”
  • Lookups. “Show me the top 20 customers by revenue last quarter.”
  • Explanations. “Why does this dataset flag tenure as the strongest predictor?”
  • Follow-ups. Conversations are stateful — refer back to “that chart” or “the previous result” and Summand carries the context forward.

​
How it works

Every response is grounded in your dataset, not a generic LLM guess:
  1. Summand picks the right tool — SQL execution, semantic-layer lookup, or chart rendering.
  2. Generated SQL is shown alongside the answer so you can audit it.
  3. Results stream in as the model reasons; long queries don’t block the UI.
  4. Charts render inline and can be exported to a report.

​
Where to find it

Open any dataset and switch to the Ask Summand view. The chat panel sits next to your data, and a full conversation history is saved per dataset.

​
Access

Ask Summand respects dataset permissions: you can only query datasets you own or have been shared on. The same chat is available to AI clients via the MCP server, so Claude Desktop, ChatGPT, and Cursor can ask the same questions on your behalf.