Run tests
EARLY ACCESSRun test groups with multiple test cases in sequence to validate agent behavior at scale.
NOTE
For quick manual testing and debugging, use Preview directly in the AI agent builder. Preview lets you send messages and view responses in real time. Quality Center is for running bulk predefined test cases.
Run a test group
Run the test cases in a test group to validate agent behavior. You can run test cases multiple times to ensure consistency and reliability.
- Open the test group you want to run.
- Select Run test group.
- In the window that opens, configure the following:
- AI agent version: Specify which agent version you want to test.
- Number of runs: Specify how many times to run the test group. You can specify a maximum of 10 runs. The default value is 1.
- Run tests multiple times to ensure that the agent responses are consistent, reliable, and accurate across different runs.
- Select Run test. This runs all the test cases in the test group. While the test runs, a progress indicator shows the current run, for example: 1/5 runs.
- When testing completes, select Go to test run to view the results.
Past runs
The Past runs tab displays all test executions for a test group.
On this page, you can do the following:
- Search: Search for test executions by using the search box.
- View test executions: View a list of test executions. For each test execution, you can view the following information:
| Column | Description |
|---|---|
| Date | Date and time when the test execution was run. |
| AI agent version | Version of the agent that was tested. |
| Test case results | Summary showing the percentage, number of passed test cases, and number of failed test cases. For example: 70%, 7 Pass, 2 Fail. For tests that are running, this shows In progress. |
Next steps
- Review test results: After running tests, review the results to identify issues and verify agent behavior.