Most of the features I regret shipping started the same way. I had an idea I was sure about, I built it, and almost nobody used it. I had been building from my own head instead of from what people were actually doing.
User interviews are the cheapest fix for that. A user interview is a focused conversation where you learn how someone works, what they struggle with, and what they have already tried. Run them well and you save weeks of building the wrong thing. Run them badly and you walk away with false confidence and ship anyway.
This is how I run them while building Peerlist.
A user interview is for learning what people already do. It is not a place to pitch your idea. The moment you start selling, the answers stop being useful.
Why this matters more now with AI
Building used to be the hard part. It is not anymore. You can describe an app to an AI tool and have something working by the end of the afternoon.
When building is cheap, the limiting factor is no longer whether you can build it. It is whether anyone needs it. I see a lot of builders, myself included on bad weeks, making the thing that is fun to build instead of the thing someone is waiting for. AI makes that trap easier to fall into, because it removes the friction that used to force you to stop and ask if the idea was any good.
A polished demo built in a weekend feels like proof. It only proves you can build, which was never the question. The best ideas still come from noticing a real problem someone has, which is the whole argument of Paul Graham's How to Get Startup Ideas. The conversation with the person who has the problem is the part AI cannot do for you (yet!), and it is the part most people skip.
Top comments (0)