AI Trace Logs help you understand how an AI-generated response was produced by showing the interpreted customer request, knowledge sources used, tools invoked, agent information, and execution details. You can use AI Trace Logs to review how an AI Agent generated a response in a ticket or chat, identify which knowledge sources were referenced, determine whether skills or tools were used, and review response latency.

This article explains how to enable trace logging, view traces from tickets and chats, and understand the information displayed in the AI Trace dialog.

Prerequisites

Before using AI Trace Logs:

• An AI 2.0 Agent must be configured in the Agent Builder portal.
• The Agent Builder permission is required to configure AI Agents and enable Trace Logging.
• Trace Logging must be enabled for the AI Agent.
• The AI-generated response must have been generated after Trace Logging was enabled.
• You must have access to the ticket or chat conversation containing the AI-generated response.

What AI Trace Logs Capture

Benefits of Using AI Trace Logs

For support agents

• Transparency: Review the sources, tools, and execution details behind AI-generated responses.
• Validation: Verify that responses were generated using the expected knowledge sources.
• Troubleshooting: Identify whether a library, skill, API tool, or MCP tool was used when generating a response.

For administrators

• Auditing: Review how AI-generated responses were produced.
• Configuration Validation: Confirm that agents are using the expected libraries, skills, and tools.
• Performance Analysis: Review response latency and execution details.

How to Enable Trace Logging for an AI Agent

Trace logging is configured per agent. Enable it before the agent goes live so that every production response has a trace.

1. Go to AI module > Agents and open the AI agent you want to trace.

2. Open the agent’s Settings and go to Trace Logging.

3. Turn on Trace Logging.

4. Select the details to capture:

   • LLM Tracing
   • Tool Tracing

5. Click Save.
   

   
   
   

How to View AI Trace Logs in a Ticket

1. Open the ticket.

2. Navigate to the AI-generated comment in the conversation.

3. Locate the AI indicator on the AI-generated comment.

4. Click the AI indicator.

5. The How AI generated this answer dialog opens.
   

   
   
   

How to View AI Trace Logs in a Chat

1. Open the chat conversation and find the AI-generated message.

2. Click the More options (⋯) menu on the message.

3. Select AI Activity Logs.

4. The AI Trace dialog opens and shows the complete trace for that message.
   

   
   Iiwib3JnaWQiOiIzIiwiaXNzIjoic3VwcG9ydC5ib2xkZGVzay5jb20ifQ.aGZTqpTRhSVbKSR4ajZW1LdVzmueO2BQslgXS_7Zrgk)
   

Understanding the AI Trace Dialog

The AI Trace dialog contains three primary sections:

• Response Summary
• Sources AI Used
• Detail Tabs

Response Summary

The header section provides a summary of the AI-generated response.
It includes:

• AI Agent avatar
• Response source summary
• Response latency
• Response timestamp

Examples:

• Answered from your knowledge base

• Answered in 11.9s

• Oct 16, 2026 03:55 PM
  

  
  
  

The exact values vary depending on how the response was generated.

Sources AI Used

This summary shows what the AI Agent relied on to generate the response. All items are clickable links—click any library, skill, or tool to view its details.

Detail Tabs

The lower section of the dialog provides additional information through two tabs:

• Technical Details
• Execution Data

Technical Details

The Technical Details tab provides information about the AI Agent used to generate the response.

Execution Data

The Execution Data tab provides detailed information about how the AI Agent processed the request and generated the response.
The tab supports two views:

Steps View
The Steps view provides a visual timeline of the response-generation process.
Common information displayed includes:

• The execution component (for example, AGENT or MEMORY).

• The operation performed, such as:

  • memory.history.load
  • agent.initialize
  • agent.invoke
  • memory.history.save

• The duration of each step.

• The overall execution timeline.

  
  
  

This view helps identify which stages of the request took the most time during response generation.

JSON View
The JSON view displays the underlying trace data captured for the response.
Depending on the trace, this information can include:

• Trace identifiers

• Session identifiers

• Message identifiers

• Start and end timestamps

• Total execution duration

• Execution status

• Metadata

• Additional trace attributes
  

  
  
  

Use this view when reviewing detailed execution information or troubleshooting AI response generation.

Troubleshooting

AI Trace Is Not Available for a Response

Possible causes include:

• Trace Logging was disabled when the response was generated.
• The response was generated before Trace Logging was enabled.
• The response was not generated by an AI Agent.

Some Trace Sections Are Empty

This behavior is expected when a library, skill, API tool, or MCP tool was not used during response generation.
For example, the trace may display messages such as:

No library sources used for this response.
No skills executed for this response.
No API tools used for this response.
No MCP tools used for this response. 

Frequently Asked Questions

1. Why can’t I see a trace for an AI-generated response?
   AI Trace Logs are available only for responses generated after Trace Logging was enabled for the AI Agent.

2. Can I view trace logs for older responses?
   No. Responses generated before Trace Logging was enabled do not have trace information available.

3. Why does the “Understood As” text look different from the customer’s original message?
   The Understood As field displays the interpreted request used by the AI Agent during processing. It may differ from the customer’s original wording.

4. Why does “Library Used” show “No library sources used for this response”?
   This indicates that the AI Agent generated the response without using a knowledge library during that interaction.

5. Why are Skills, API Tools, or MCP Tools empty?
   These sections display only the resources used during response generation. If a resource was not required, the trace indicates that no corresponding skills or tools were used.

6. Does enabling Trace Logging create traces for past responses?
   No. Trace Logging applies only to responses generated after the feature is enabled.

7. Are AI Trace Logs available in both tickets and chats?
   Yes. AI Trace Logs can be viewed for supported AI-generated responses in both tickets and chat conversations.