I Spent Months Trying to "Revive" Our E2E Tests. Now I'm Building My Own AI Tool.

​My name is Ilya, and I'm a QA Engineer.

​To be exact, I'm the only QA on my team. You know what that means. I handle regression, new features, talking to devs, and explaining "why this broke again."

​But this isn't a story about starting automation from scratch. This is a story about inheriting it.

​The "Gift" I Got Tired of Fixing

​When I joined the team, I was given a "gift": a big, heavy, and complex automation project.

​"Great," I thought, "at least something is already here. I'll just clean it up."

​Oh, how wrong I was.

​I spent the first few months not testing, but just trying to get this project to run. I updated libraries, rewrote entire modules, and fought with the environment.

​And you know the worst part? Even after all that work, it was still unstable. The main problem was flaky tests.

​Every "red" build in our CI turned into a detective story: is this a real bug, or did a selector just break because the page changed by one pixel?

​Spoiler: 9 times out of 10, it wasn't a real bug. It was just a brittle test.

​I got tired of fixing. My job is to break the product, not to fix the tests.

​"What if...?" (The Idea)

​One day, looking at another "red" build, I realized: as the only QA, I can't afford to spend my time maintaining this brittle system.

​I sat there thinking. We live in an age where AI can write code and explain complex topics.

​Why can't I just ask it:

​"Hey AI. Go to mysite.com/login. Click the 'Login' button. Type '[email protected]' in the 'Email' field. Check that you see 'Welcome' on the page."

​Why am I hunting for selectors by hand if an AI could "see" the page like a human and adapt to changes?

​Of course, it's not that simple. A "general" AI (like in a chatbot) knows nothing about the DOM, what "click" means, or my specific website.

​But what if I could take those powerful AI APIs and build a "bridge"? Create a service that translates my plain English into actions a browser can understand?

​So, I decided to build it myself

​And that's how Debuggo.app was born.

​For the last 3 months, I've been building this tool alone, in my spare time after my day job.

​Let me be honest: I'm not building my own neural network. I'm a solo developer, not a giant corporation.

​My job is to be an integrator. I take powerful, "general" AI APIs and build smart logic around them. I'm teaching them to understand the context of web testing.

​I'm teaching them how to find buttons even if the id changed, and how to type text into fields even if the layout is tricky.

​My goal is to build a tool that lets me (and you) write automated tests as fast as we write test cases in a notepad.

​Why am I writing this? (I need your help)

​I'm building Debuggo because it's the tool I desperately need in my daily job. I'm building it for myself and for people like me—QAs (or devs) who never have enough time to maintain monster frameworks.

​But I can't build it in a vacuum. As the only QA, I only know my problems. I need to see how it works on your websites.

​I'm opening up the beta.

​If you know the pain of "flaky" tests. If you're also tired of maintaining an old automation project. If you're willing to try something new (and maybe a little rough around the edges)—I would love to get your feedback.

​Check out https://debuggo.app and sign up for the beta.

​Let's try to automate this automation. Together.