Autopilot logs record every action taken by the AI, including plan generation, assignment selection, messaging, and monitoring decisions.
Each log entry includes a timestamp, a description of the action, the type of entry (e.g. analysis, plan creation, assignment, message), the AI’s justification when applicable, and the status (pending, approved, rejected, completed).
Logs provide an audit trail so you can understand why the AI recommended a certain activity or message and ensure it aligns with your clinical judgement.
Accessing Client Logs
On the Autopilot assignment page, locate the client whose logs you wish to view and click the logs icon/link (if available). Alternatively, you can navigate directly to the logs page using the URL parameters for the client and team.
The logs page is organized by Autopilot job. A job represents one complete run of the Autopilot—analysis, plan generation, assignments and messages.
Reading the Logs
Each job header lists the job number, the completion date and time, and how many log entries are contained within. Click a header to expand or collapse its entries.
Within an expanded job, each row shows the time of the action, what the AI did, the category of the log, and the AI’s reasoning (if provided). The status column indicates whether the action is pending approval, has been approved, has been rejected, or has completed automatically.
For messages and assignments that require approval, you will see a “pending” status. Approve or reject these items from your normal workflow; the logs page is read‑only and serves as a record.
If the AI was unable to generate a plan, assignment, or message, the reasoning will note that no valid response was produced. You may need to adjust the Autopilot settings or provide additional context.
Using Logs to Improve Autopilot
Review logs regularly to see how well Autopilot’s recommendations align with your expectations. If you notice patterns you dislike, adjust the therapist style, categories, outcomes or monitoring settings and save a new preset.
Compare the AI’s justifications to your own clinical reasoning. Use this information to refine the custom instructions or provide additional client context.
Keep track of outcomes by matching log entries with client progress. This helps you evaluate the effectiveness of automated interventions and make data‑driven decisions.