Traditional testing approaches are slowing down development and are no longer scalable. We believed that test automation was the solution. This belief failed. We test using test automation tools almost everywhere these days, but this testing approach is still too slow and requires strong technical skills that are hard to find and develop. Even if we do succeed in automating our testing, we are then faced with exhausting maintenance. Is there a better solution? Could it be autonomous testing? Autonomous testing is the next phase in the evolution of testing approaches. However, the market is full of promises, plenty of test automation tools with tiny AI-based features, and extremely high expectations of what this topic could do. I will share an overview of where the market currently is and compare reality with expectations. In this talk, I will give you a few tips on how to build your own testing bot. You will need to be able to collect data, learn some (often AI-based) advanced algorithms for interaction with the system under test, and later also similarly for validation/assertions. All with the usage of open-source tools. We didn’t find a compact autonomous testing tool, so we decided to not wait anymore and begin with several experiments to build our own autonomous bot. Finally, I will also show you our first results from the implementation of our autonomous visual regression testing.