Activity Overview
The Activity page shows all conversations in a two-panel layout:- Left panel — List of conversations with filters
- Right panel — Detailed view of the selected conversation
Conversation List
The left panel displays all conversations with:| Info | Description |
|---|---|
| Conversation ID | Unique identifier for the conversation |
| Preview | First message from the user |
| Timestamp | When the conversation occurred |
| Status | Success, error, or other states |
Filtering Conversations
Use filters to find specific conversations:Date Range
Select a start and end date to view conversations within a time period.Status
Filter by conversation outcome:- Success — Conversation completed normally
- Error — An error occurred during the conversation
Message Count
Filter by number of messages in the conversation.Search
Search for specific text within conversations.Conversation Details
When you select a conversation, the right panel shows three tabs:Chat Tab
View the conversation as it appeared to the user:- User messages (right side)
- Agent responses (left side)
- Timestamps for each message
- Tool calls and their results
- Agent’s reasoning process (when applicable)
Details Tab
Metadata about the conversation:| Field | Description |
|---|---|
| Conversation ID | Unique identifier |
| Start time | When the conversation began |
| End time | When the conversation ended |
| Message count | Total messages exchanged |
| Token usage | Tokens consumed |
| Model | AI model used |
| Status | Final conversation status |
| Cost | Estimated cost (if enabled) |
Traces Tab
Technical debugging information:- Raw trace data from the AI model
- Detailed execution timeline
- Useful for debugging complex issues
Understanding Tool Calls
In the Chat tab, you’ll see when your agent used tools:- Which tool was called
- What parameters were sent
- What response was received
Common Patterns to Look For
Successful Conversations
- User asks a question
- Agent retrieves relevant information (from knowledge base or tools)
- Agent provides a helpful response
- User is satisfied
Tool Usage Issues
- Agent attempts to use a tool
- Tool returns an error
- Agent may try alternative approaches or apologize
Knowledge Gaps
- User asks about something not in your knowledge base
- Agent indicates it doesn’t have that information
- Opportunity to add content to your knowledge base
Using Activity for Improvement
Identify Common Questions
Look for patterns in user questions:- Frequently asked questions → Add to knowledge base
- Repeated tool usage → Ensure tools are working well
- Common confusion → Improve system prompt
Spot Issues Early
Monitor for:- Error conversations
- Tool failures
- Unhelpful responses
Measure Performance
Track over time:- Conversation volume
- Success rates
- Tool usage patterns
- Token consumption
Exporting Data
Click the download button to export conversation data for further analysis.Privacy Considerations
Conversation data may contain:- User questions and context
- Personal information shared by users
- Tool responses with customer data