Hey HN,<p>together with a friend we realized in our engineering jobs that e2e testing is often broken and not worth the effort. But AI Agents could change this massively.<p>There are a couple of reasons why AI agents could revive the topic:<p>- LLMs can automatically adjust tests to the evolving product by simply looking at the git diff and changes in the UI<p>- It'll become much easier and cheaper to have functioning E2E tests<p>Trends supporting this:<p>- Vibe coding is on the rise and the increasing amount of changes needs to be tested<p>- Software will be interactive, like games, where the task and suggestion comes to the user instead of the user navigating to the workflow<p>Are current tools just not good enough yet, or do people simply not care enough about the problem?