The Observability section provides platform users with comprehensive visibility into agent operations, enabling data-driven insights through session analysis and trace monitoring.
Sessions
Track and analyze individual user interactions with your agent:
Session History: View complete conversation logs for each session
Key Metrics (filterable by time period):
Average session time
Total sessions
Average traces per session
Average token consumption
Traces
Understand the technical flow of agent interactions by examining how messages are processed:
Trace Details: Visualize the complete journey of each query including:
LLM processing approach
Triggered tasks and tools
Step-by-step execution summary
Performance Insights (filterable by time period):
Total traces
Average span per trace
Average latency
Average token consumption
Both traces and sessions serve different purposes. Sessions just provides the interaction overview between user and agent. Whereas traces focus on summarizing how bot is responding to user query for every conversation pair.
Both sections include time-based filters, allowing you to analyze trends and patterns across different periods for better optimization and troubleshooting.
