Smart Verify checkpoints

Supported on versions 26.3 and later, Relevant for: GUI tests and components

Smart Verify checkpoints let you check anything you want about an object's text or visual appearance using generative AI.

Overview

In a Smart Verify checkpoint, you describe what you want to verify in plain language, without going into technical detail. For example, you could have a step that verifies the validity and relevance of a date range or email address, or a step that makes sure all items displayed on the screen match a selected filter.

The checkpoint passes if the application meets the expected conditions. Otherwise, a step failure is reported in the run results, along with a reasoning explanation from the LLM.

The Smart Verify results and reasoning are provided in English, but you can use any language your LLM understands to describe the condition to verify.

The following table describes how smart verifications differ, depending on the type of object you are checking.

Checkpoint attribute On property-based test objects On AI Objects
When do you create the checkpoint?  During a recording session During AI Object Inspection or AI Record sessions
Are checkpoints stored in the object repository? Yes

No

All checkpoint properties are reflected in the step in the editor and can be edited there.

What can I ask questions about? 

 The object's properties or image.

The object's text or image.
What object method represents this checkpoint? 

The Check method.

For details, see the Object Model Reference.

The Smart Verify method.

For details, see the Object Model Reference.

Prerequisites

To create and run Smart Verify checkpoints, you must have the following prerequisites: 

  • A valid UFT Aviator add-on license. For details, see License editions.

  • A configured LLM connection. For details, see UFT Aviator setup.

    Note:  

    • If necessary, you can create a checkpoint before configuring the LLM connection, but you will not be able to test the validity of your verification statement or run the step until you configure the connection.

    • If a Smart verify step cannot be performed during the test run for lack of license or LLM connection, the step is marked as failed and the test continues.

Create or edit a Smart Verify checkpoint

You can create Smart verify checkpoints during a recording session, an AI Record session, or an AI Object Inspection session.

To define Smart Verify Checkpoint Properties:

  1. Open the Smart Verify Checkpoint Properties dialog box, in one of the following ways: 

  2. For property-based object checkpoints: 

    Accept or modify the checkpoint name, which, by default, is the same as the object name.

    The object's class is displayed in read-only mode.

  3. Select whether to check the object's text, properties, or visual appearance (image).

    Verification type Description
    Object properties (for property-based objects) or Object text (for AI objects)

    Verify that an object or its text meets specified characteristics.

    Object image

    Verify that an object's visual appearance matches your expectations.

  4. Enter a plain-language verification statement that describes what you want to verify about the object. For example, "The button is enabled and displays the text Submit", or "The error icon is displayed in the top right corner".

  5. (Optional) Click Test to test your verification statement against the application object you selected for the checkpoint.

    Tip: Limit image-based verification to small objects, to reduce the use of LLM tokens and to avoid the connection timing out before the complete image can be transferred.

    The test returns a True or False indication. Click the drop-down arrow by the result to view the LLM's explanation for the result.

  6. Click Add to test to save your changes or add the Smart Verify step to your test.

See also