The initial customer is still hard

Building is getting easier. What remains hard is selling it to someone.

The first user is always me. That part is solved: I know what I want, and AI can put a working system in front of me in days. The problem starts with the second person, and the third. How do I pitch what I built to someone who is not me?

Part of it is that AI builds instant systems, but people don’t live in instant systems. They live in a relatively constant environment and grow along with it. Nobody wants to start from scratch, even when starting from scratch is now free. So the problem is no longer “Can I build this?” It is how to match what I build to the desire of one individual, one niche need, and then the next.

I think about it in architectural terms. An architect delivers the building. It is finished, and it is empty. What makes it someone’s home is the interior: how the rooms are arranged, the furniture, where the light falls. AI has turned a single person into a developer who can put up an entire apartment building alone. But the customer still needs an interior of their own, and the job now is to convince them to pay for one.

Maybe Base44 was the first to make this happen. One person, no funding, and Wix bought it within its first year for eighty million dollars. I went to look at how it is doing now: two million people building on it, a hundred and fifty million dollars a year.

Base44 industrialized the shell. I was shocked when I first encountered it, but the bottleneck is still quite clear: to build something more than a toy, the shell that Base44 provides is not enough. For a business, it falls short; for a toy, it is easily discarded. It’s like a Russian nesting doll (matryoshka / 套娃) that never settles into a finished shape.

As for selling it: can we hack it through the platform feed? How can regular people gain attention without big platforms or media monopolies?