If using BQL generation feature during limited availability, it is highly recommended to enable only the relevant Subject Area Entities to improve performance.
- Sign in to the Birst space in which to enable generative AI features.
- Select . On the Data Model Description screen, add entities to expose specific attributes and measures.
- Add descriptive labels and descriptions to each column.
Labels and descriptions improve Insights summaries even if attributes or measures are not enabled for BQL generation.
- On the Embedding Processing screen, select attributes whose unique values to embed in AI models.
This helps AI models choose appropriate filter values in BQL queries.
- On the Instructions screen, specify additional information and context about the data model.
Include information that helps the AI model generate accurate BQL queries and add any domain-specific knowledge relevant to your data.
- Click Prompt Playground in the top toolbar to test BQL query generation with natural language prompts.
If required, adjust your configuration based on the test results.