Yes, one person with AI can now do the work that used to take ten, or fifty, or a hundred. I've done a version of it. But the conclusion a lot of people draw from that, fewer people and smaller teams, is the wrong one, and it's wrong in a way that bites.
AI collapsed the build. The operation is another matter, and it didn't collapse with it.
The operation doesn't collapse
Building the architecture, one operator with AI is a genuine step-change, mostly because the communication overhead just disappears. Once the thing is live, you're somewhere else. There's incidents to handle and the slow, unglamorous maintenance that never makes a headline. None of it runs on one person, because a person sleeps, eats, goes to the shops.
We haven't had an incident at Kash. But it's a real risk and I treat it like one. I've built runbooks, feature flags and kill switches, enough that someone non-technical can step in and stop the bleeding while I'm asleep. What they can't do is fix the root cause. That waits for me to wake up and charge my phone. Picture a vulnerability with money at risk at three in the morning, and the only person who can actually fix it unreachable. That isn't a one-off you design around. It's the standing condition of a single-technical-person system, and AI doesn't touch it. Bugs come either way.
The headcount maths is backwards
So "one person can build it, so we need fewer" has the economics backwards. If AI hands each person a step-change in output, banking that as a headcount cut is the bearish move. Same output, smaller team, and you've capped yourself. The better play is to keep the people and turn the new capacity into more. More revenue, more of the markets that never penciled out before. If you can't find more to create with the extra capacity, something's saturated or broken upstream.
Who comes out ahead
The people who win are the ones who actually embrace this and get good at it. Management thins out. Leadership doesn't, because someone still has to point twenty agents at the right problem and keep the humans around them motivated. It'll be harder for a junior to land the first rung, which I hate, because it's already brutal out there. I've mentored a few people fresh out of university and they're scared. I think it's a short-term retraction, and it'll settle.
The value is moving around. Small companies can produce what only big ones could before. Big ones can lift their throughput too. The danger sits in the middle, squeezed from both ends, while a lot of new ground opens up that the old cost base never allowed.
So if you're reading the solo-build story as a reason to shrink, you've got it backwards. AI made the building cheap. Running it safely, and answering at 3am when money's on the line, still takes people.