For most of the last two decades, building a software product required enough technical skill that a large share of bad ideas quietly died before they ever reached a customer. Not because the idea was tested and failed, but because nobody with the ability to build it thought it was worth the months of work. That filter was crude and unfair, but it was still a filter.
AI-assisted development is removing that filter faster than any new filter is arriving to replace it. That's mostly a good thing, a lot of genuinely good ideas died in exactly that bottleneck, from people who understood a problem deeply but couldn't build. But it creates a specific new failure mode worth naming honestly: the amateur founder problem.
What the problem actually is
An amateur founder, in this specific sense, is not someone who is new to business. It's someone who can now produce a working, polished-looking product without ever having been forced to develop the judgment that used to get built, almost as a side effect, during the long and expensive process of building something the hard way. Under the old constraints, spending six months building a product was also six months of being forced to talk to users, argue about scope, and confront whether the idea was actually worth the effort, because the cost of being wrong was so high.
Remove that cost, and you can arrive at a working product without ever having done any of that thinking. The product looks the same. The judgment behind it is not.
Why this matters strategically, not just personally
This isn't only a risk for individual founders. It changes what a full market of new entrants looks like. If the friction that used to force a certain minimum of validation and judgment onto every founder disappears, the average quality of judgment behind new products should be expected to fall, even while the average technical polish of those products rises. Investors, partners, and early customers evaluating a new product can no longer treat technical execution as much of a signal of the founder's judgment, the two used to be correlated by necessity, and that correlation is breaking.
Where this leaves me
I don't think the answer is to be nostalgic for the old friction, it was a genuinely bad filter that excluded a lot of good ideas from people who couldn't code. The more useful response is to notice explicitly that a signal the market used to get for free, technical execution as a proxy for founder judgment, is disappearing, and to ask what should replace it. My current guess is that direct evidence of contact with real users and real constraints, rather than the polish of the product itself, has to do more of that signaling work than it used to.