Independent Health leveraging AI

Mobile app poses many testing challenges

When Independent Health announced a new mobile app as an easy way for members to track their deductibles, review benefits, and view claims, most people were excited, including Chris Trimper, enterprise QA architect for Independent Health. However, along with the launch of this new mobile app came an exponential increase in new test maintenance responsibilities. “Don’t get me wrong, I thought it was a great idea and could absolutely see how it would benefit our members. But, from my application testing perspective, it posed a big challenge. Testing for two operating systems, Android and iOS, with different implementations, locators, and object identifiers had the potential to significantly increase our test maintenance cost and effort. I was also worried that we would not be able to keep up with the fast pace of change in a mobile app. Our UX and UI colleagues were planning a regular release schedule of small changes, often stylistic, amounting to an overwhelming amount of test automation work.

“People don’t realize that something simple such as changing the color of a button, moving it to a different location on the page, or giving it rounded edges, requires lots of ‘underground’ code changes and therefore test changes. I’m all for innovation and helping our members, but I just wasn’t sure how best to ensure it meets our quality standards within the required timeframe.”

Increased test automation efficiency with AI capability

A long-time OpenText™ user, Independent Health already leverages OpenText™ Application Quality Management (ALM/ Quality Center) for all manual and automated test management. This is used in conjunction with ALM Lab Manager and seamlessly integrates with Azure DevOps. All application tests are triggered from OpenText Application Quality Management and processed through OpenText Functional Testing. The team uses OpenText™ Professional Performance Engineering (LoadRunner Professional) for all web, REST, and SOA protocol performance testing.

Once Trimper found out that OpenText™ has also been implementing Artificial Intelligence (AI) test capabilities within OpenText Functional Testing, he had to learn more about it. OpenText Functional Testing’s new AI-based capabilities increase automation efficiency by simplifying and improving test creation, execution, and maintenance. By infusing AI in the process, objects are identified in the same way a human does. Because AI understands each object and the contextual interaction, a single test script can execute tests on multiple platforms.

“We were skeptical at first, but after a demonstration, it was like OpenText™ had handed us an ‘easy’ button,” says Trimper. “We could see straight away how this would save us an immense amount of time. automated back-end code-based testing and AI user level testing can complement each other perfectly. With a growing AI icon library, OpenText Functional Testing (UFT One) experiences the application just like a user would. We save time by testing only the actual components that a user interacts with on the screen, and our tests look more like a user story, focusing on workflow and business requirements.”

OpenText Functional Testing AI capability unifies test automation in diverse application landscape

Following the successful launch of the first mobile app, Independent Health continues to leverage OpenText Functional Testing for its automated testing effort in all business-critical mobile and web applications, web services, and databases such as Oracle, PostGres, and Microsoft SQL Server. Most web applications are responsive and integrate with other solutions, so business workflows start in one system and move to a different one, while ensuring a backend database record is updated. End-to-end testing is of vital importance. A test scenario could simulate the life of an Independent Health member through enrollment, accessing the customer portal, making changes to their benefit plan, submitting a claim, following the claim processing, and receiving claim payment.

“Our application landscape has developed over a number of years,” explains Trimper. “Under the cover of each of the applications we use different programming languages. For instance, we have a major .NET application that we UI test with OpenText Functional Testing (UFT One). We use lengthy or short tests depending on the scenario we’re working in, but we drive the interface in exactly the same way an end user would. We only need to modify the UI test scripts when the developers change the flow of the application to alter the user experience, or when they add new features. This is a huge time saver for us. We also work a lot with hybrid applications, for instance a Java application that uses a browser for display. When you look at it, it’s a website, but it has different underlying technology and some controls that aren’t common in a web browser. No problem at all for OpenText Functional Testing (UFT One), the web add-in ensures that objects from different technology worlds can live side by side. We test our database connections using OpenText Functional Testing (UFT One) API tests which give us user-friendly drag-and-drop capabilities and a defined test workflow.”

OpenText Functional Testing Lab for Mobile and Web (UFT Digital Lab) reduces test maintenance by 35%

The team uses OpenText Functional Testing Lab for Mobile and Web for android and iOS application testing, as Trimper explains: “We realized AI adoption could really revolutionize our mobile testing effort. Instead of having to write and maintain two sets of test scripts and two sets of object repositories, we can develop these against just one device.”

With OpenText Functional Testing Lab for Mobile and Web’s (UFT Digital Lab) AI-based testing capabilities, we can leverage a single set of multiplatform scripts across iOS and Android and reduce mobile test maintenance by at least 35 percent. This makes us much more responsive to the fast-changing nature of mobile apps. We don’t need to use a costly device to run all our testing either; instead, we use device emulators and save costs.

Independent Health also realized it could introduce test creation earlier. Even if the lay-out development isn’t quite finished yet, the team can create an AI test case based on a mockup or screenshots; ready to execute when the code is finished. AI-based location and app interaction means that nearly all locator fails disappear. As much of test maintenance is centered around locator issues, this significantly reduces maintenance efforts. The underlying platform really doesn’t matter as the test automation acts like a user.

The way forward with ‘automagic’ AI-based test creation

The team has enjoyed AI in the context of OpenText Functional Testing add-on capabilities too. The PDF addon with OpenText Functional Testing wasn’t working great with Independent Health’s systems as they used letters generated by 3rd party template engines, which weren’t recognized by the addon. However, AI came to the rescue and now OpenText Functional Testing can read a PDF file exactly like a user would. This supports the compliance effort as PDF files can now be tested to ensure they contain the correct fonts and icons.

“You need to make the mental shift from assuming that test automation needs to be complicated, when with AI it is just simple and easy, completely focused on the user experience,” says Trimper. “AI can teach us about the application too. Sometimes we run the AI component to determine how a user interacts with the app and give us ideas for further enhancements.”

OpenText Functional Testing’s (UFT One) AI-based capabilities simplify the tough bits from test creation. I’ve known testers struggle for hours to interface with a particular visual element. AI takes all this away, and I can’t imagine spending more than a few minutes on the same thing. I call it ‘automagic’ as it really is!

When the AI-plug in for OpenText Functional Testing is activated, there is a ‘screenshot always on’ option that Trimper encourages his development team to use: “This gives you a visual narrative of everything that you’re doing. Because sometimes the problem is not what’s on the screen right now, but it’s the screen before or the interface interaction you just did. Having the visual history gives you an easy tool to go back and check without having to rerun everything.”

Reduce stress and workload for specialized testing teams

Trimper can see how AI can support a DevOps testing environment more effectively: “This is still such a growth area. We’re getting closer to natural language testing which will offer up testing to a wider audience and reduce the stress on specialized testing teams. Every new OpenText Functional Testing (UFT One) release gives me something exciting to help reduce our time-to-market and fully support the evolution of our new mobile app.”

He concludes: “We collaborated in a spectacular partnership with OpenText™ on the development of the OpenText Functional Testing’s (UFT One) AI-based capabilities. OpenText Functional Testing (UFT One) and OpenText Functional Testing Lab for Mobile and Web (UFT Digital Lab) combined now support our automated testing efforts across mobile, and web applications, covering the core of our application infrastructure. Having multi-platform and multi-device test ability, without needing platform development expertise, has revolutionized the way in which we support our members.”