There is still no substitute for a founder personally using their own app

Automated browser testing handles bugs, but it doesn't replace the judgment of a human founder using their own product. The smartest models in the world still lack an obvious kind of common-sense intelligence — the ability to recognize when something is clearly wrong in a way that any user would notice. Right now, the right approach is to personally go through your entire app every so often, and then quickly spin up agent sessions to fix any bugs you find. We're still far from agents that can replicate this kind of intuitive product sense.