When I'd Still Hire

I’ve become a fan of building alone. So let me argue against myself for a minute, because a take that can’t say where it breaks isn’t a take, it’s a pitch.

Here’s where solo plus AI breaks.

Scale, first. Zero to one is cheap alone. What comes after product-market fit is not: reliability, on-call, security, compliance, the boring infinite maintenance tail. That’s where teams earn back their coordination cost, and pretending otherwise is how solo founders end up with a successful product and a pager that never stops. Success is precisely the thing solo doesn’t absorb well. The reward for winning a bet is a workload that no longer fits in one person’s week.

Continuity, second. One person is one point of failure. I can get sick, get bored, get hit by the proverbial bus, or get absorbed by a different product for a month. If the thing matters and needs to outlive my attention, a team is partly insurance, and insurance is worth paying for. Nobody thinks about insurance while the sun is out, which is exactly when you have to buy it.

Then the hard, integrated, known problems. Deep infrastructure, regulated systems, anything where the moat is doing one difficult thing extremely well at depth. That’s concentration-of-force territory. The portfolio logic that makes solo powerful for discovery is precisely wrong here: you don’t want five cheap attempts at a flight-control system. Hire.

There are also domains where a second pair of eyes is a safety requirement, not a nicety. Payments, health data, anything where a mistake lands on someone else. Self-imposed AI review isn’t the same thing, and I wouldn’t trust myself to pretend it is. The value of a human reviewer was never only the review. It was that they didn’t share my blind spots, and every agent I direct inherits at least some of them.

And there’s the relational work. Enterprise sales, partnerships, the kind of trust that gets built over dinners and years. Humans, not agents, and not one human doing everything. I learned that selling companies, not building them.

The synthesis is a phase distinction, not a verdict. Solo plus AI wins discovery. Teams win scale and depth. The skill is knowing which phase you’re in and not bringing the wrong structure to it. Most people fail in one direction: hiring too early, paying the coordination tax before they know what they’re building, shipping their org chart before they have anything worth an org. But staying solo too long is a real failure too, just a quieter one. Nobody writes the postmortem for the product that stalled because its one builder wouldn’t let go.

The default flipped, but a default isn’t a rule. Build alone until the thing you’re building tells you it’s time not to. The part I watch for now, and I don’t have a clean answer, is whether I’ll actually hear it when it does. The same independence that makes solo work feel so good is a bias against noticing the moment it stops working.