AI-Assisted Testing Reloaded: What’s New – and What the Future Holds

We introduced our AI-assisted testing concept and solution at our workshop last October, and it has generated significant interest in the market ever since; we have already launched several such projects in the enterprise sector.

What exactly is AI-assisted testing? In a nutshell, our solution is built on a workflow consisting of AI agents that support testers starting from the requirements analysis phase. It is capable of creating standardized, detailed test cases that can be easily executed by less experienced testers and easily automated with the help of either test engineers or agents responsible for test automation.

The solution can be readily adapted to various enterprise systems, and thanks to its use, the time spent writing test cases can be reduced by as much as 30–50%, thereby significantly reducing the entire testing process.

For a bit more detail, it’s worth reading our previous blog post on the topic.

AI-Assisted Testing Reloaded: What’s New…

More than six months have passed since last October—what happened in the meantime? Our experts have added three important new features to the system:

1. Graphical User Interface

The user interface makes the system easier to use: users managing the system can upload documents (primarily specifications) here, based on which the system generates test cases. Individual functions listed in the specifications for which test cases can be requested can also be specified—this way, it is not necessary to generate test cases based on the entire document, saving time.

2. New AI Agent: Test Coverage Reviewer Agent

The task of this new agent, which reinforces the team of test case-generating agents, is to review the completed test cases to ensure they cover every branch of the functions defined in the specification. If necessary, it issues instructions via a prompt to the agents responsible for creating test cases to generate new ones. This further improves test coverage and, ultimately, the quality of the finished software.

3. Parallel Generation of Test Cases

The system is now capable of generating test cases for multiple functions simultaneously. This increases its performance and reduces the total time required for testing. While processing a 150-page specification previously took 6–7 hours, thanks to parallel processing, this time has been reduced to one hour.

…and What The (Near) Future Holds

Of course, development never stops. Let me briefly introduce three ongoing improvements below.

1. Workflow Integration into SDLC

In early 2026, we launched our AI-assisted SDLC solution, which covers the entire software development lifecycle, including testing. It stands to reason that the testing phase of this solution should be the subject of this article—our AI-assisted testing solution, which is already complete—and we are currently putting the finishing touches on this integration.

2. Test Suite Rationalization

It is particularly common in large organizations for the test suite to swell to thousands of tests; this is especially true for regression testing. The reason for this is that successive testing teams create more and more test cases, and due to a lack of time, reviewing these, comparing them with existing ones, and filtering out overlapping test cases rarely happens. This results in testers frequently executing test cases that are very similar to one another, which is an unnecessary and time-wasting solution.

Thanks to the duplicate filter feature, however, the system can highlight such overlapping test cases, allowing the test manager to easily decide whether to merge or delete them—thereby minimizing duplicate testing and reducing the time spent on test execution.

3. Static Frontend Testing

During static frontend testing, the testing agent first reviews the application design document (e.g., Figma) and then examines the frontend interface of the completed application. It compares the two and indicates whether the completed frontend conforms to the specifications.

Conclusion

As we stated earlier, our goal is not to build an “off-the-shelf” product. We are developing the expertise and methodology to offer flexible solutions tailored to the needs of a given organization, which can then be adapted to any large-scale enterprise process or system.

 

Is it also typical for your company that you struggle to meet testing deadlines, and the quality of the finished software has room for improvement? We should talk.