---
title: "How to react when an AI quality assessment is ready"
description: "Start a Scenario Builder section when AI QA finishes a dialog assessment — without polling the assessments table."
---

# How to react when an AI quality assessment is ready?

When a dialog closes, AI quality assessment still runs **afterwards**. A section that starts from chat closed often sees an empty score and summary. The **AI QA assessment completed** event fires when the assessment is actually ready.

## When you need it

- Send the score and summary to CRM after the dialog is closed.
- Call a webhook or write Custom Data when AI QA has finished.
- Replace a temporary workaround: polling assessments from a custom action.

## What to know

- The event fires only after a **successful** AI assessment. A failed assessment does not start the section.
- The section can read `trigger_data.score`, `trigger_data.dialog_summary`, `trigger_data.chat_room_id`, `trigger_data.assessment_id`.
- A second successful AI evaluation of the same dialog starts the section again.

## Before you start

- [x] AI QA is enabled and a checklist has an agent.
- [x] You are signed in as an integrator or administrator.
- [x] You have a section with a Start node.

## Steps

1. Open the section in Scenario Builder and select the Start node.
2. Set the trigger type to **Internal events**.
3. Choose **AI QA assessment completed**.
4. Add a send node (CRM, webhook, Custom Data) and map fields from `trigger_data`.
5. Save and publish the scenario.

Field details: [Internal events trigger reference](/en/scenariobuilder/reference/internal-events-trigger-reference.md).

## How to verify

1. Close a test dialog that AI QA will assess.
2. Wait until the assessment status is completed in Quality Assurance.
3. Confirm the section ran and the external system received the score.

## Typical mistakes

| Problem | Fix |
|---------|-----|
| Section runs but the score is empty | Use **AI QA assessment completed**, not chat closed. |
| Section does not run | The assessment may have failed. This event ignores failures. |
| Duplicate send | A second AI evaluation fires the event again. Deduplicate in CRM or in the scenario. |

## Related

- [Why quality assessment arrives after chat close](/en/scenariobuilder/explanation/why-qa-assessment-completed-event.md)
- [How to configure internal events for the Start node](/en/scenariobuilder/how-to/configure-start-node-internal-events.md)
- [Internal events trigger reference](/en/scenariobuilder/reference/internal-events-trigger-reference.md)