Sessions

Sessions group related traces together, making it easier to monitor and debug complex workflows in your LLM applications. By organizing traces into sessions, you can track the entire lifecycle of a user interaction or multi-step process.

​

Understanding Sessions

A session represents a logical grouping of related traces. Common examples include: Conversations - All messages in a chat with a specific user. Workflows - Multi-step processes like document analysis or data processing. Batch Jobs - Related operations processing multiple items. User Journeys - Complete user interactions from start to finish.

​

Session Components

Each session includes: Session ID - Unique identifier that links traces together. Session Name - Human-readable name for easy identification. Traces - All prompt executions associated with the session. Timeline - Chronological view of all activity in the session. Aggregated Metrics - Total costs, tokens, and durations across all traces.

​

Viewing Sessions

Access sessions in the AgentMark dashboard:

1. Navigate to the Sessions tab
2. Browse or search for specific sessions
3. Click a session to view:
   • All traces in the session
   • Timeline of events
   • Total token usage and costs
   • Custom metadata
   • Performance metrics

​

Best Practices

Use Consistent IDs - Ensure all related traces use the same session ID. Meaningful Names - Give sessions descriptive names for easy identification. Include Context - Add relevant metadata to help with debugging and analysis. Limit Scope - Keep sessions focused on specific workflows or interactions. Session Lifecycle - For long-running applications, create new sessions periodically rather than endless sessions. Clean Separation - Don’t mix unrelated operations in the same session.

​

Integration

To enable sessions in your application, include session information in your telemetry metadata when executing prompts. See the Development Observability documentation for implementation details.

​

Next Steps