Generate and review AI reports
Turn a Vision task's recordings into an AI report, read its issues and fixes, mark which are real, and see what it cost.
An AI analysis report reads all the finished recordings of one test task and explains what stopped your testers. It gives the task a UX score, lists the issues with their root cause and a suggested fix, and ranks the fixes. You then confirm the issues that are real and send them to your tracker.
Before you begin
- To generate a report or review its issues, you need the Editor, Admin or Owner role. Every role can read reports.
- The task has an approved happy path and at least one finished tester recording.
- Your organization has tokens. Each report spends tokens from the shared balance. See Tokens.
Generate a report
Open the task's reports
In the sidebar, click Flamio Vision > Test Tasks, click the task, then click View analysis. The AI analysis reports page opens.
Check that a slot is free
The badge beside the generate button shows how many of your reports are running and how many you may run at once. If the button reads Limit reached, wait for one of them to finish.
Start the report
Click Generate report. If Generate reports asynchronously is on in Settings > Analysis, the button reads Generate async report instead.
If your balance is zero, the Out of tokens dialog opens instead of a run. Click Upgrade plan to top up.
Wait for it to finish
A progress bar shows the run. A standard report finishes in minutes. An async report can take up to 24 hours — Flamio confirms with "Batch submitted. Up to 24h."
Result: the report appears in the list with its date, the number of sessions analyzed, its UX score, Issues, Risk and Status. When Status reads Completed, click the row to open it. The Feedback column shows Sent once you have reviewed it.
Read a report
The report opens as Optimization Report, with the number of sessions analyzed and the date at the top. It has these parts:
| Section | What it tells you |
|---|---|
| Executive Summary | The UX score out of 100, the risk level, Path Deviation (how far testers strayed from the happy path), Issues Found, Critical, and a short summary. |
| Expected User Flow | The approved steps of the happy path, with the number of issues found at each step. |
| Issues | Each issue groups what testers did — wrong choices, hesitation, dead clicks, rage clicks, going off-script, abandoning — under one root cause, with how many testers it affected and the steps it touched. |
| Issues Found | When a report has no grouped issues, each finding is listed with its severity, the problem, why it matters, the Root cause and a Suggested Fix. |
| Alternative Paths | Ways testers went that differ from the happy path but still completed the task. |
| Fix Priority | The fixes in order, with Priority, Problem, Fix and Impact. |
To see the evidence for an issue, click Watch Recording — or the recordings count when several recordings show it. A side panel plays the few seconds around the moment and links to the full replay.
Review the issues
You need the Editor, Admin or Owner role.
Mark each issue
Click Confirm on issues that are real and Reject on those that are not. When you reject one, a box asks Why is this not a real issue? — your answer is optional.
Add your overall feedback
Under Your Feedback, enter a note about the report as a whole.
Save
Click Save Feedback. The section shows how many issues you confirmed and rejected.
After you save, click Export approved at the top of the report to create tickets from the confirmed issues. The button works once a task tracker is connected to the project in Settings > Integrations. See Connect an issue tracker.
See what a report cost
Every role can see this.
Open usage
Click Settings in the sidebar, then Usage under Billing.
Find the report
Under Usage history, the Activity column reads AI analysis or AI analysis (async) for a report, with the project and the tokens it spent.
To spend less per report, or to ask for deeper analysis, change the options in Settings > Analysis. See Project settings.