We have built an AI agent that runs tests for any platform using only natural language and visual recognition.
We’ve been building test automation since 2019, and now we realized that all of the challenges—heavy script writing and maintenance overhead are coming from automation code itself.
So what if E2E testing didn’t require code at all?
Instead of generating or maintaining test code, Aximo:
1. Takes a goal in natural language 2. Observes the UI visually 3. Plans and executes actions 4. Evaluates results via vision 5. Iterates until success or failure
It doesn’t rely on DOM selectors or frameworks and observes the interface visually and decides what to do next — similar to how a human tester interacts with software.
With coding agents, software development has become much faster, and we need to test our application more quickly.
We’re curious what this community thinks.