Insights and Analytics
Analytics
AI agents

AI agents analytics

EARLY ACCESS

AI agents analytics enables you to monitor performance and usage metrics for your AI agents and tools. Track agent effectiveness, response times, interaction volumes, and tool utilization to optimize your AI-powered customer service operations.

AI agents analytics consists of two main categories:

  1. AI agent performance - Monitor AI agent performance, tool usage, and session activity. Access specialized metrics through three dedicated tabs: AI agents, Tools, and Sessions.
  2. Knowledge agent analytics - Track knowledge base performance and content effectiveness. Analyze how AI agents utilize your knowledge resources to provide accurate responses.

AI agent performance

AI agent performance provides comprehensive insights into your AI agents, their tool usage, and session activity. Monitor performance metrics, identify optimization opportunities, and track how effectively your AI agents handle customer interactions.

At the top of the page, you can switch between the following tabs:

  1. AI agents
  2. Tools
  3. Sessions

AI agents

The AI agents tab displays performance metrics for all your AI agents. Monitor agent effectiveness, response times, and interaction volumes to ensure optimal performance.

Summary

MetricDescription
Total agentsTotal number of AI agents in your system.
Average response timeAverage time taken by AI agents to respond to customer interactions (in milliseconds).
Total interactionsTotal number of interactions handled by all AI agents.
Total failuresTotal number of failed AI agent interactions.

Performance by agent

The Performance by agent section provides detailed metrics for each AI agent:

ColumnDescription
Agent nameName of the AI agent.
InteractionsNumber of interactions handled by this agent.
Average response time (ms)Average response time for this agent in milliseconds.
FailuresNumber of failed interactions for this agent.
Success ratePercentage of successful interactions for this agent.

Tools

The Tools tab displays performance metrics for all tools used by your AI agents. Monitor tool effectiveness, call volumes, and success rates to ensure your AI agents have access to reliable and efficient tools.

Summary

MetricDescription
Total toolsTotal number of tools available to AI agents.
Total tool callsTotal number of times tools have been called by AI agents.
Average call timeAverage time taken to execute tool calls (in milliseconds).
Overall success ratePercentage of successful tool calls across all tools.

Performance by tool

The Performance by tool section provides detailed metrics for each tool:

ColumnDescription
Tool nameName of the tool.
Total callsNumber of times this tool has been called.
Average call time (ms)Average execution time for this tool in milliseconds.
Failed callsNumber of failed calls for this tool.
Success ratePercentage of successful calls for this tool.

Sessions

The Sessions tab displays a table of all AI agent sessions. Use this tab to review individual session details and track interactions.

ColumnDescription
Session IDUnique identifier for the session.
AI agentThe AI agent that handled the session.
End userThe end user who participated in the session.
DateDate of the session.

Knowledge agent analytics

Knowledge agent analytics enables you to monitor performance of your knowledge agents and identify opportunities for improvement. Track how effectively your AI agents utilize knowledge resources, analyze response patterns, and optimize documentation based on real usage insights.

There are two task types:

  • Analyze user traffic - Analyze user messages and knowledge agent responses.
  • Evaluate - Evaluate knowledge agent performance.

Analyze user traffic

Analyze messages your users sent in the selected date range and get insights into knowledge agent performance.

The report provides a detailed list of user messages, the agent response for each message, and the context on which the answer is based so you can better understand why it answered a certain way. It also generates topics.

Important

You can analyze up to 4,000 messages. If more messages exist in the selected period, a random sample is taken across the entire period.

Create task

To create a task:

  1. Select a knowledge agent from the drop-down menu.
  2. Enter task name.
  3. Select either chatbots or senders from the drop-down menu to filter the analysis. You can only select either chatbots or senders, not both.
  4. Choose the From and To date.
  5. Select Run task.

After you run the task, it appears with the details in the task table below.

Evaluate

Evaluate tests your knowledge agent with premade questions. It provides a list of all agent responses for those questions and the context on which each answer is based, so you can understand why the agent answered a certain way.

To run an evaluation, upload a .csv file that contains only one column with test questions for the knowledge agent. These questions should be questions your users are likely to ask when interacting with your business. You can download a CSV example to see the expected format.

Create task

To create a task:

  1. Select a knowledge agent.
  2. Enter task name.
  3. Select Run task.

Tasks

The task table keeps a history of the last 20 tasks per task type.

ColumnDescription
Task nameName of the task.
Knowledge agentThe knowledge agent that was analyzed.
Created atDate and time when the task was created.
StatusCurrent status of the task (In progress, Failed, or Completed).

You can view a detailed report or download it for each completed task.

Detailed report

You can access the detailed report of a task by selecting the first icon in the Actions column of the Tasks table. You can also select the download icon to save it to your PC.

This opens a page with two tabs: Overview and Insights.

Overview

The Overview tab displays the following summary metrics:

MetricDescription
Total interactionsTotal number of interactions analyzed.
Avg response timeAverage time the AI assistant took to respond.
Success ratePercentage of successful responses.
Documents retrievedNumber of different documents used to generate responses.
Avg interactions per sessionAverage number of interactions in each session.
Avg interactions per userAverage number of interactions per user.

The Overview tab also includes:

  • Insight distribution - Breakdown of different insight categories.
  • Response latency - Hourly latency trends.

Insights

The Insights tab categorizes interactions into the following types:

CategoryDescription
Successful answersInteractions where the AI assistant provided correct and helpful responses.
Improve documentationInteractions where the documentation could be improved to provide better answers.
Not documentedInteractions about topics that are not covered in the documentation.
Unexpected messagesInteractions with unexpected or off-topic user messages.
Unusual examplesEdge cases and unusual interaction patterns.

Select a category tab to view the individual interactions that fall under that category.



Need assistance

Explore Infobip Tutorials

Encountering issues

Contact our support

What's new? Check out

Release Notes

Unsure about a term? See

Glossary
Service status

Copyright @ 2006-2026 Infobip ltd.

Service Terms & ConditionsPrivacy policyTerms of use