Prompt Playground
The Prompt Playground is a testing tool that administrators use to validate a GenAI configuration before AI capabilities are available to end users. The tool displays how the AI generates BQL queries based on the current configuration of Data Model Descriptions, Embedding Processing, and Instructions.
Access the Prompt Playground from the GenAI Settings page. Click the button in the upper-right corner of the page.
Configuring Data Model Descriptions, embeddings, and instructions is an iterative process. The Prompt Playground helps you test configuration changes and evaluate results before users access AI capabilities.
Use the Prompt Playground to:
- Verify that the AI selects the correct measures and attributes for a question.
- Confirm that embeddings match attribute filter values correctly.
- Verify that custom instructions are applied correctly.
- Identify configuration gaps, such as missing labels, descriptions, or embeddings that produce inaccurate queries.
- Compare results before and after configuration changes.
The Prompt Playground panel contains:
| Element | Purpose |
|---|---|
| Prompt | A text field where you type a natural language question |
| Subject Area selector | Limits BQL generation to selected subject areas, using only GenAI-enabled measures and attributes to improve relevance and accuracy. Falls back to the Default Subject Area if none is selected. |
| Generate Query | Button that sends your prompt to the AI and returns the generated BQL |
| Clear | Resets the prompt and response fields |
| Response | Displays the BQL query the AI generated for your prompt |