AI Agents statistics

⚠️ Required permissions: the AI Agents Statistics permission must be enabled in your permission model. To understand how permissions affect the data you can access, see Manage access to statistics.

⚠️ Prerequisite: the AI Agents feature must be enabled on your account. If it is not enabled, the AI Agents tab will not appear in Statistics, even if your profile includes the required permission.

The AI Agents tab, available from Statistics, lets you track how learners use AI agents. It is divided into three sub-tabs: Discussions, Learners and Simulations.

Choose a sub-tab

Discussions

The Discussions sub-tab displays one row per conversation. The table can include the following columns:

• Alias;
• Email;
• Discussion topic;
• Reported;
• Report reason;
• Report comment;
• Agent;
• Discussion started;
• Time spent;
• Activity result.

Click View to open the conversation in read-only mode.

The Open discussion column is available only if your profile includes the Log in as a user permission. The Open button then lets you access the conversation in the app as the learner.

Learners

The Learners sub-tab groups data by learner. It includes:

• Alias;
• Email;
• Total discussions;
• Time spent;
• Last used;
• one column showing the number of discussions for each type of agent.

By default, only the Simulation and Recommendation agent columns are displayed. Use Columns to add the other agent columns, as well as First used.

Click a learner’s row to view their discussions.

Simulations

The Simulations sub-tab groups data by simulation. The table can include:

• Sessions;
• Learners;
• Time spent;
• Completion rate;
• Success rate;
• Average score;
• Reported;
• First used;
• Last used.

Only sessions linked to a saved simulation are included. If a simulation has been deleted, its historical data remains available under Deleted simulation.

Click a row to view the learners who took part in that simulation. This view also includes the Best score indicator.

Identify which agent was used

In the Discussions sub-tab, the Agent column identifies the type of agent used. Four values can appear:

• Simulation;
• Q&A;
• Consolidation;
• Recommendation.

These correspond to the agents configured under Select available AI agents:

• Simulation Agent;
• Question Agent;
• Consolidation Agent;
• Recommendation Agent.

The Activity result indicator can display Passed, Failed or In progress. It is populated only for discussions linked to a simulation.

💡 An empty Activity result cell for a Q&A, Consolidation or Recommendation discussion is therefore expected.

Read simulation results

The Simulations sub-tab lets you track completion, success and scores across simulation sessions.

⚠️ Average score is expressed as a percentage. The % symbol is not displayed: an Average score of 62 therefore means 62%.

In the simulations table, it corresponds to the average score across evaluated sessions. In a learner’s details, it corresponds to the average score across that learner’s own evaluated sessions.

⚠️ Completion rate and Success rate are calculated from all sessions started. A session that was started and then abandoned is therefore included in both calculations.

A session that did not generate an evaluation is counted under Sessions, but is not included in the Average score calculation. It is therefore normal for the total number of sessions to be higher than the number of sessions contributing to the score.

Identify a learner

The Alias column displays the alias chosen by the learner. If no alias has been set, the learner’s name is displayed instead.

⚠️ The information shown in the Alias column can therefore vary from one learner to another. To identify a learner more reliably, use the Email column.

Understand reported discussions

The Reported column indicates whether the learner reported an agent response from the app.

When a discussion has been reported:

• Report reason shows the reason selected by the learner;
• Report comment shows any additional comment they provided.

These fields remain empty when a discussion has not been reported.

In the Simulations sub-tab, the Reported indicator shows the number of sessions associated with at least one report.

Add an internal comment

The Internal comment column lets you add a private note to a discussion. Click the pencil icon in the cell to add or edit the comment.

Internal comments are visible only to back-office users and can contain up to 2,000 characters.

Filter the data

The filter panel includes the main filters available across Statistics:

• Period, set to Last 90 days by default;
• Associated department(s), when the department hierarchy is enabled;
• Groups;
• Include deactivated learners;
• Advanced filters.

These filters restrict which learners and data are included in the tables. To understand exactly how they work, see Filters and audiences.

Customize the columns

Use Columns to show or hide columns in the table.

Some available data is hidden by default. Depending on the sub-tab, this can include Success rate, Reported, Best score, First used, or columns related to specific agents.

Export the data

The Excel export button generates a file based on the table currently displayed, including the visible columns.

Before exporting, use Columns to keep only the data you need.

Good to know

Test sessions are not included. Tests run from the back office while configuring a simulation do not appear in any of the three sub-tabs. They do not affect the rates or the Average score.

These statistics measure how learners use AI agents. They do not provide information about costs, consumption or the number of requests generated by AI agents.

A simulation’s score depends on how the simulation is configured. The AI evaluates the learner’s performance based on the objectives and levels defined in the simulation. The way these elements are configured therefore directly affects the Average score displayed in Statistics.

Learn more

• What are our learning agents?: discover the available AI agents, enable them and define which groups can access them.
• Configure a simulation: understand how the objectives and levels defined in a simulation affect the scores shown in Statistics.
• Configure AI: configure the tone, provider and model used by AI agents.
• Filters and audiences: understand how the period, department, groups and other filters affect the data displayed.