QUERY Mode
QUERY mode is the UI workflow for read-oriented analysis: generate SQL, inspect outputs, and export results.
What QUERY is for
Section titled “What QUERY is for”- Convert questions into SQL-oriented analysis.
- Review results as Table, Chart, or Raw output.
- Export data for reporting and sharing.
Before you send a QUERY request
Section titled “Before you send a QUERY request”- Attach relevant schema context in Metadata Explorer or use Add Selection from a script with related metadata.
- Set the output format to Table, Chart, or Raw.
- Provide a clear prompt for the analysis.
End-to-end QUERY workflow
Section titled “End-to-end QUERY workflow”- Select QUERY mode.
- Choose the output format.
- Optionally choose a chart type.
- Send the prompt.
- Review the result and export CSV or PNG when needed.
1) Preflight clarification
Section titled “1) Preflight clarification”Preflight now always runs for QUERY requests to confirm table, field, filter, and grouping intent before execution.
- Automatic: Preflight runs by default for every QUERY prompt.
- Clarification: If Genie detects ambiguity, it asks short questions to ensure the generated SQL is correct.
- Budget: You can control the maximum number of clarification questions in Settings > Clarification Budget > Preflight Questions.
2) Output format selection
Section titled “2) Output format selection”Choose between Table, Chart, and Raw output directly from the composer.

3) Chart type selection
Section titled “3) Chart type selection”When Chart is selected, choose a chart type before sending.

4) Result views
Section titled “4) Result views”Chart result view
Section titled “Chart result view”
Raw data result view
Section titled “Raw data result view”
Exports
Section titled “Exports”- Table output: Export CSV from the result block.
- Chart output: Copy PNG or Export PNG from chart actions.

Accuracy tips
Section titled “Accuracy tips”- Keep schema selections tightly scoped to the request.
- Use explicit table and field names for critical reports.
- If results are broad, refine the prompt and rerun.
Best practices
Section titled “Best practices”- Validate complex logic in QUERY before moving to DATA execution.
- Prefer smaller iterative prompts for large or complex questions.
- Save exports with clear names for traceability.