What is a Session?
A session groups related traces together. When your AI system processes a user conversation or runs a multi-step workflow, all the traces from that interaction share the same session ID. Sessions give reviewers the full context they need. Instead of seeing isolated traces, they see the complete interaction.Why Sessions Matter
| Without Sessions | With Sessions |
|---|---|
| See individual messages | See entire conversation |
| Missing context | Full context for decisions |
| Hard to understand intent | Clear picture of user needs |
How Sessions Work
Sessions are created automatically by Velatir integrations. When you use the browser extension, n8n node, LangChain middleware, or SDKs, traces are automatically grouped into sessions based on the interaction context. For example:- A chatbot conversation creates a session with all messages
- An n8n workflow creates a session for all steps in that run
- A LangChain agent creates a session for the entire agent execution
Session Status
Sessions have a status based on their traces:| Status | Description |
|---|---|
| Active | Recent activity in the session |
| Awaiting Action | Has traces with pending review tasks |
| Inactive | No recent activity |
Viewing Sessions
In the Velatir dashboard:- Go to Sessions to see all sessions
- Each session shows:
- Number of traces
- First and last trace timestamps
- Current status
- Sources involved
- Click a session to see all its traces in order
Session Metadata
Each session displays:| Field | Description |
|---|---|
| Trace Count | Number of traces in the session |
| First Trace | When the session started |
| Last Trace | Most recent activity |
| Status | Active, Awaiting Action, or Inactive |
| Sources | Systems that contributed traces |
Best Practices
Group Logically
A session should represent one logical interaction:- One user conversation
- One workflow execution
- One batch processing run