Skip to main content

​
Chat Log

Use the Chat log tab to inspect individual conversations handled by an agent. It’s invaluable for QA, support follow-up, and training improvements.

​
Getting Started

  1. Open Insights and select an agent.
  2. Choose a date range at the top of the Chat log tab (e.g., Today, Last 7 Days, custom).
  3. Scroll the transcript list to view interactions chronologically.
If there are no conversations in the selected window, the page displays “No data”.

​
Transcript Details

Each entry includes:
  • Timestamp: When the conversation began.
  • Channel: Chat, voice, embed, etc.
  • Outcome Flags: Indicators such as unattended sessions or escalations.
  • Preview Snippet: First few lines of the conversation for quick scanning.
Click an entry to open the full transcript within the panel.

​
Export Options

Export Conversations

Downloads every transcript in the filtered range—ideal for archival or in-depth analysis.

Export Unattended

Generates a report containing only sessions where the agent didn’t resolve the user’s request. Useful for training targets.
Exports typically download as CSV or JSON (depending on your workspace configuration).

​
Best Practices

  • Weekly QA Reviews: Have subject-matter experts read a sample of transcripts to ensure quality stays high.
  • Tag Knowledge Gaps: Annotate sessions where the agent lacked information, then add sources via Add Content.
  • Monitor Unattended Chats: Frequent unattended conversations suggest tuning the agent instructions or connecting additional tools.
  • Share Feedback: Distribute notable transcripts to product and content teams to highlight wins or issues.

​
Troubleshooting

  • Missing Conversation: Expand the date range or confirm the user spoke with the correct agent.
  • Large Exports Timing Out: Narrow the window (e.g., export a week at a time) or upgrade your plan’s export limits.
  • Privacy Concerns: Mask sensitive data before sharing exports externally, or use the Privacy tab to limit access.
The Chat log tab keeps every exchange transparent, giving you the evidence you need to iterate on prompts, workflows, and knowledge sources.