Self-serve data Q&A spaces for a business team
Set up a curated question space over a few tables with instructions and sample queries so a team can ask its own data questions.
The use case
Analysts field the same ad hoc questions from sales or ops every week. Databricks Genie lets a data team build a space for one audience: pick the relevant tables, write instructions about business terms, and add example SQL. Business users then ask questions in plain language and get tables and charts, with the generated SQL visible. The data team monitors questions that failed or got thumbs-down and improves the instructions. Users still check surprising results with an analyst, and access is limited by Unity Catalog permissions.
Example workflow
- Pick tables and the audience for the space
- Write instructions and example queries
- Invite a pilot group
- Review failed or flagged questions weekly
- Refine instructions and widen access
Tools
- Databricks Genie
- Unity Catalog
Risks and guardrails
- Too many tables lower accuracy; keep spaces focused
- Users may trust wrong answers; show the SQL
- Permissions must be set at the catalog level
Sources
Written by Agents Lookup. Credits and official pages for the tools this use case names:
- Databricks docs: Genie: docs.databricks.com/aws/en/genie
- Databricks Genie (official page): docs.databricks.com/aws/en/agents/mcp/genie-mcp (checked 2026-11-19)