Running Autonomous Test Execution

Run a manual test case automatically with Autonomous Test Execution in Test Manager, using an AI model, an execution target, optional live streaming, and Orchestrator credential references.

Context

This task runs an existing manual test case automatically using Autopilot, instead of executing it by hand or building a scripted automation.

Tip:

Autonomous Test Execution can also be launched from a Pending Execution at the test case log level.

Run stays disabled if the selected runtime type and machine template don't match, or if no connected runtime exists for the selected runtime type. The execution folder is still resolved automatically to the project's default folder — there's no folder picker on the Configure tab. Configure tab selections aren't persisted between sessions. Reopening the panel resets the following selections to their defaults:

• Runtime type
• Robot account
• Machine template
• Host machine

Note:

Manual steps written for a human tester may not be unambiguous enough for the agent. Wording may need adjusting based on where a run fails before it executes reliably.

Prerequisites

• Test Manager must be delivered through Test Cloud. Autonomous Test Execution is not available in Test Manager delivered through Automation Cloud, Automation Cloud Public Sector, or Automation Cloud Dedicated. See the Test Manager feature availability page.
• A connected runtime of one of the supported types (Testing, App Testing, Production (Unattended), Cloud – VM Testing, or Cloud – VM) in Orchestrator, in the same tenant, with access to the project's default folder.
• The AssetsView permission in Orchestrator, if the test case references credential assets.

Steps

1. Log in to Test Manager.
2. Open a manual test case.
3. Go to the Manual Steps tab and select Run autonomously in the right-hand sidebar. The panel opens on the Setup tab.
4. Enter the target application URL for the web application under test. The field is not pre-filled with the test case value, and the Run button stays disabled until a URL is entered. The run uses the default browser configured on the machine (Chromium-based browsers and Firefox are supported), and browser selection is not configurable.
5. Add parameters or credentials as needed: enter {{ in a step description to trigger autocomplete for parameters, or reference a credential asset with {{credential:credentialName}}.
6. Select an artificial intelligence (AI) model.
7. Optional: Enable live streaming.
8. Optional: Select the Configure tab and, under Runtime type, select one of Testing, App Testing, Production (Unattended), Cloud – VM Testing, or Cloud – VM. The two serverless runtime types are not available for Autonomous Test Execution. A runtime type with no connected runtime appears disabled. A runtime is pre-selected automatically when one is available, and can be overridden.
9. Under Robot account, select a robot account. The default value is Any User/Robot account.
10. Under Machine template, select Any machine template or a specific machine template. Selecting a specific machine template makes the Host machine field available.
11. Under Host machine, select Any connected machine or a specific machine.
12. Select Run.

Result

Autopilot executes the manual test case. While the run is in progress, it can be watched live, remote control can be taken if needed, and the run can be canceled. Once the run finishes, either successfully or with errors, Back to Setup becomes available to reconfigure and start a new run without closing the panel.