Session Tracing Data Model for observability
Key Takeaway
Agentforce Session Tracing ties all of an agent session's data together under one session ID. In the Session Tracing Data Model (STDM), AIAgentSession is the top-level container, and AIAgentInteraction is one turn. AIAgentInteractionStep records each action or operation the agent performed, and it is the main layer for troubleshooting.
The Session Tracing Data Model (STDM) is a set of Data 360 objects that records what an Agentforce agent did during each session, from the user's first message to the agent's final response. You can build reports, dashboards, and queries on it to see agent behavior, debug problems, and find ways to improve the agent. Agent Analytics and Agent Optimization are both built on top of the STDM.
What Agentforce Session Tracing captures
Agentforce Session Tracing captures each agent session from start to finish. It pulls data from different sources, starting with reasoning engine logs, and ties all of it together under one session ID.
The log for a session shows:
- Turn-by-turn interactions
- Reasoning engine executions
- Actions
- Prompt and gateway inputs and outputs
- Error messages
- Final responses
You can use the trace data to test, debug, and monitor agents in sandbox or production environments.