Chat with your data

Ask a question

From the home page, open a new conversation and type your question in plain language — no SQL required. Describe what you want to know, for example "how many orders were shipped last month?". The assistant looks at the data sources you have access to and turns your question into a query.

You can start a chat directly from the home page, from a specific database or dataset in the data catalog, or from a file you have uploaded. Starting from a data source gives the assistant more context, which usually means more accurate answers.

Context & follow-ups

A conversation keeps your context. You can refine a result by asking follow-ups such as "now only show the European region" or "aggregate this by week". The assistant remembers the previous query and adapts it instead of starting from scratch.

You can also reference tables and columns explicitly in your question. If a term is ambiguous, the assistant may ask for clarification or show you which data it used before running anything.

Tips for good questions

• Be specific about time ranges, regions, or metrics ("last 30 days", "per country").
• Name the metric or business concept you care about (revenue, churn, active users).
• Mention the data source when it matters ("in the sales database").
• For complex requests, break them into smaller follow-up questions.

The clearer the question, the more reliable the generated SQL. When in doubt, ask a simple question first and refine from there — it is often faster than writing one long prompt.