UFT Aviator

Supported on versions 26.3 and later

UFT Aviator is an embedded AI assistant, providing support for generative AI capabilities.

Overview

Different types of AI-based features are available: 

Feature type Description
AI Object Detection

AI-based testing identifies objects in your application as a human would, and interacts with them by performing device-level operations where the objects are visible.

You can describe objects in your script manually, use Inspection to learn them from your application, or create steps using an AI Record session. AI objects are not stored in the object repository. 

For details, see AI-based testing.
Generative AI features

Generative AI features extend beyond the AI Object Detection capabilities and require a UFT Aviator license.

UFT Aviator connects to your own Large Language Model (LLM) and can provide features where you describe tasks in plain language.

For details, see Smart verification.

Note: When responses are AI-generated, there is no guarantee as to the accuracy, completeness, or appropriateness of the information. Make sure to apply your judgment, fact check, and consider multiple sources before making any decisions.

Prerequisites

Before using Aviator for Generative AI features, review the following prerequisites.

Prerequisite Details
License

You must have a valid UFT Aviator license. For details, see License editions.

In the product user interface, a lock icon near a feature option indicates that you do not have a license available.

Setup Configure the connection to your LLM as described in UFT Aviator setup.

UFT Aviator setup

This section describes how to configure UFT Aviator to work with your preferred LLM.

Only models based on the OpenAI protocol with chat completion endpoints are supported. Refer to the Support Matrix for a list of supported models.

To configure the connection to the LLM:

  1. In the AI title bar menu, select UFT Aviator Configuration.

  2. In the LLM Provider list, select OpenAI, Google, Mistral, or Custom.

  3. Enter the provider base URL, the API Key you use to access it, and the exact Model name.

    For the provider base URL, enter only the API base path up to and including the API version. Do not include the model name, endpoint path, or operation suffix. For example, enter: https://generativelanguage.googleapis.com/v1 or https://api.openai.com/v1

  4. Click Test connection, to make sure OpenText Functional Testing can access your LLM.

  5. Click Save to store the configuration you entered, then close the UFT Aviator Configuration dialog box.

Note:  

  • To generate AOM script steps for the configured LLM settings, open the Options dialog box, in the GUI Testing > General pane, and click Generate Script. See also Sample Generated Script for Global Testing Options.

  • The API key is stored in an encoded format, both in the stored configuration and in AOM scripts.

  • Clicking the Restore Factory Defaults button in the Options dialog box resets all product settings, including the UFT Aviator Configuration settings.

Smart verification

After you set up UFT Aviator to communicate with your LLM, you can use generative AI to check anything you want about your objects. Specify the object, select whether you want to check the object's properties or something about it's visual appearance, and describe what you want to verify in plain language.

You can perform smart verifications in your tests using one of the following:

  • Smart Verify checkpoints on property-based test objects. You can create these during a recording session and they are stored in the object repository.

  • Smart Verify steps on AI Objects. You can create these during AI Object Inspection or AI Record sessions, and they are not stored in the object repository.

For details, see Smart Verify checkpoints.

See also