eval-session-classify

Session intent satisfaction is classified by analyzing a session_id alongside Datadog LLM Observability traces and RUM signals across pre-, during-, and post-event windows to produce a structured satisfaction verdict.

Automated session classification verdicts are generated by integrating ROUTE_CONTEXT details, LLM Observability evaluations, and RUM signals to robustly measure whether user intent was satisfied.

The classification pipeline validates the presence of required data sources and handles missing LLM observability traces safely to ensure robust satisfaction verdicts without processing failures.

Yes, you can provide a standalone session_id to measure whether user intent was satisfied by integrating RUM signals and LLM Observability traces across the session timeline.

Pre-, during-, and post-event data windows are required to capture the complete session context, ensuring the classification pipeline produces robust and accurate satisfaction verdicts.