How to verify AI text responses after column data type changes in Bold BI v15.3 and latest
Bold BI v15.3.8+ introduces an AI improvement specifically for text-based responses generated by the AI Assistant and AI Agent. According to the official release history, this version enhances how the AI handles schema and data type mismatches when column data types are modified.
This article explains how to verify that text-based AI responses correctly reflect updated column data types after changes in the underlying data model or query.
Step-by-Step Instructions
1. Confirm your Bold BI Server is running v15.3.8 or later. If you are on an earlier version, consider Upgrading to Bold BI Server v15.3.8+
2 Navigate to the Datasource designer
3. Change the column data type in your underlying data source (or query) used by the Bold BI dataset.
Example: Number → String, String → Number, or Number → Date.
4. Click on Save button to modify the column.
5. Reopen the dashboard that uses the dataset and ask the AI Assistant/AI Agent a question that uses the modified column.
Example: “Show sum of Salary” (if a column changed from Number to String) or “What is the average Salary?” (if changed to string).
6. Confirm the AI text response matches the updated data type and does not show conversion errors.
Expected result on v15.3.8+:
- The response reflects the updated column type
- No “cannot convert …” or similar schema-mismatch failures appear
For best results, always use the actual column names as defined in the dataset when prompting the AI Assistant or AI Agent.
Notes
- This improvement applies to text-based AI responses (AI Assistant/AI Agent).
- The improved behavior applies only to actions performed after upgrading to v15.3.8.
- Existing dashboards typically do not require changes after upgrade, but you must refresh the dataset after schema changes to ensure Bold BI picks up the new metadata.
Related Articles:
