The Generative AI Audit Trail provides audit records from Collibra AI requests generated within Collibra DQ.
When a request is sent to the AI engine, all events from the request and response are logged on this page. This allows Collibra DQ administrators to see who used Collibra AI to generate or troubleshoot SQL and when, the input prompt and suggested response query, and how long it took for Collibra AI to generate a suggested query.
You can also export the audit trail records to CSV, Excel, or PDF, print them, or copy them to the clipboard.
The following table describes the columns included on the table of the Generative AI Audit Trail page.
| Column | Description |
|---|---|
| Activity ID | The internal ID assigned to each event. |
| User | The username of the user who submitted the request. |
| Dataset | The name of the dataset from which the request was submitted. |
| Generation Time | The amount of time that elapsed between the time of the request and the response. |
| Page |
The page in Collibra DQ from which the request was sent. /alert signifies that the request originates from the Alert Builder page. /dq/explorer2 signifies that the request originates from the Explorer page. /dq/finding signifies that the request originates from the Findings page. /dq/profile signifies that the request originates from the Profile page. /dq/rule signifies that the request originates from the Rule Workbench. |
| Prompt | The exact content of the request sent to the Collibra AI engine. |
| Response | The exact response from the Collibra AI engine. |
| Error | The contents of the error message when the Collibra AI engine returns an error. |
| Audit TS | The date and time of the submission of the request. |