Feedback
Review AI Agent and Agent Assist Feedback
In this article, you'll learn how to review feedback on AI Agent responses and Agent Assist AI Answers including how to filter results, interpret summary metrics, and use feedback details to identify response gaps.
How it works
Feedback in Performance Insights is split into two types:
Message Feedback — ratings submitted by end users on AI Agent messages.
Agent Assist Feedback — ratings submitted by human agents on AI Answers surfaced in Agent Assist.
Each type has its own detail view with filters, summary metrics, and a CSV export.
Before you start
You must have access to Insights in your workspace.
Message Feedback requires at least one AI Agent with feedback collection enabled. To configure this, see AI Agent Message Feedback.
Agent Assist Feedback requires Agent Assist to be installed and active in your helpdesk tool.
Step 1: Open the Feedback tab
Go to Insights > Global > Performance.
Click the Feedback tab.
The Feedback tab displays two summary cards:
Message Feedback — total thumbs up/down ratings submitted by end users on AI Agent messages.
Agent Assist Feedback — total thumbs up/down ratings submitted by human agents on AI Answers surfaced in Agent Assist.
Click View Feedback on either card to open its detail view.
Step 2: Review Message Feedback
The Message Feedback view shows all end-user ratings on AI Agent responses.
Summary metrics
Metric | Description |
|---|---|
Total Feedback | Total ratings submitted in the selected period |
Positive Feedback | Count of thumbs up ratings |
Negative Feedback | Count of thumbs down ratings |
Usefulness | Percentage of ratings that were positive |
Filters
Use the top-right controls to narrow the data:
Date range — defaults to Last Three Months; filter to any supported range.
AI Agent — defaults to All AI Agents; filter to a specific agent.
Click Export to download the filtered results as a CSV.
All Feedback tab
Lists every individual feedback submission.
Column | Description |
|---|---|
AI Agent | The agent that generated the rated message |
User | The user who submitted the rating |
Feedback | Thumbs up or thumbs down |
Comment | Optional comment left with the rating |
Submitted at | Date the feedback was submitted |
💡 Sort by Negative Feedback and check the Comment column to identify specific response issues worth investigating.
Feedback Activity tab
Aggregates feedback by user, showing which users are submitting the most ratings and how their sentiment breaks down.
Column | Description |
|---|---|
User Name | Display name of the user |
Email address of the user | |
AI Agent | The agent the feedback was submitted against |
Positive Feedback | Thumbs up count from this user |
Negative Feedback | Thumbs down count from this user |
Total Feedback | Total submissions from this user |
ℹ️ A high volume of negative feedback from a single user may reflect a specific workflow gap rather than a broad agent issue. Review their conversation history in Inbox for context.
Feedback Details panel
Click any row in the All Feedback tab to open the Feedback Details panel. The panel shows:
Field | Description |
|---|---|
User Name | The user who submitted the rating |
Time | The exact date and time the feedback was submitted |
AI Agent | The agent that generated the message |
User Feedback | Thumbs up or thumbs down |
Question | The original question the user asked the AI Agent |
Answer | The full AI-generated response the user rated |
User Comment | The comment left with the rating, if provided |
💡 Use the Question and Answer fields together to evaluate whether the AI Agent's response was accurate and appropriate for the request.
Step 3: Review Agent Assist Feedback
The Agent Assist Feedback view shows ratings submitted by human agents on AI-generated answers surfaced by Agent Assist during active case work.
Summary metrics
Metric | Description |
|---|---|
Total Feedback | Total ratings submitted |
Positive | Count of thumbs up ratings |
Negative | Count of thumbs down ratings |
Filters
Use the top-right controls to narrow the data:
Feedback type — filter by All Feedback, Positive, or Negative.
AI Agent — defaults to All AI Agents; filter to a specific agent.
Time range — defaults to All Time; filter to a specific time period.
Click Export to download the filtered results as a CSV.
ℹ️ Only the filtered data is included in the CSV download.
All Feedback tab
Lists every individual Agent Assist rating.
Column | Description |
|---|---|
Case ID/Ticket | The case, work item, or incident where the AI Answer was surfaced |
User | The human agent who submitted the rating |
Feedback | Thumbs up or thumbs down |
Comment | Optional comment left with the rating |
Submitted at | Date the feedback was submitted |
ℹ️ Results are paginated. Use the page controls at the bottom right to navigate through older entries.
Feedback Details panel
Click any row in the All Feedback tab to open the Feedback Details panel. The panel shows:
Field | Description |
|---|---|
User Name | The human agent who submitted the rating |
Case ID/Ticket | The case, work item, or incident the AI Answer was surfaced on |
AI Agent | The agent that generated the answer |
User Feedback | Thumbs up or thumbs down |
AI Answer | The full AI-generated answer the agent rated, including the knowledge sources it was drawn from |
User Comment | The comment left with the rating, if provided |
💡 The AI Answer field shows the sources Enjo used to generate the response. If an answer is rated negatively, review the cited sources to identify whether the underlying knowledge needs to be updated.
What's next
Performance Insights — Review AI and human agent accuracy, feedback, and resolution gaps.
Adoption Insights — Monitor how teams are engaging with Enjo across channels and features.
AI Agent Message Feedback — Configure feedback collection on your AI Agent.