the company is getting smaller

I was in a planning conversation recently, the unglamorous kind, mapping out how to build a new product on top of an existing company. And we hit the question every plan eventually hits: how many people does this take. What’s the headcount. How do we staff it.

We both stopped, because the honest answer had quietly changed under us while we weren't paying attention. We don't need a big team anymore. We need a small number of people holding judgment, each with an enormous amount of agent leverage underneath them. I have been thinking about that moment ever since, because I think it's the shape of the next decade, and almost nobody has updated for it.

headcount was always a proxy, and the proxy is breaking

For my whole working life, the size of a company was a rough proxy for its ambition. Big thing, big team. You wanted to do more, you hired more, and the org chart grew because human attention was the scarce resource and you needed more of it to cover more surface.

That proxy is breaking. Not slowly, and not at the margins. The thing that changed is that software learned to hold a goal and reach for a tool, to do the actual work of a task rather than just assist a human doing it. Once that’s true, a single person with judgment can direct a layer of agents that covers the surface ten hires used to cover. The constraint stops being how many people you can afford and becomes how clearly the people you have can think.

I’m not speculating about this. I build this way now, daily. I direct agents in plain language and they do the work I used to do by hand, and the bottleneck in my own day moved from typing to knowing what I actually mean. When that scales from one builder to a company, the org chart inverts. Most of the boxes stop being people.

judgment becomes the whole job

Here’s the part that excites me and unsettles me in equal measure. When agents do the doing, the only thing left for the humans is the thing agents can’t yet do well: decide what’s worth doing, hold the standard, own the consequence, and know when the output is wrong in a way that matters.

That’s not a smaller job. It’s a more concentrated one. A company of five people with real agent leverage isn’t a small company doing a small thing. It’s a small number of people each operating at the scope of a department, holding judgment over a wide surface that executes itself underneath them. The skill that gets rare and valuable isn’t doing the work. It’s having taste sharp enough, and accountability real enough, to direct a lot of capable execution without it drifting into confident nonsense.

The companies that win the next decade will be small on purpose, not because they couldn’t grow, but because growth in headcount stopped being the way you get bigger. You get bigger by getting clearer.

why this lands hardest in the boring industries

The instinct is to think this future shows up first in flashy software. I think it’s the opposite. The places where agent leverage matters most are the ones drowning in necessary, unglamorous, judgment-light work that has resisted automation for decades, precisely the regulated, compliance-heavy, paperwork-dense corners of finance and operations.

That’s the lane I’m building in. Onchain markets, real-world assets, the machinery of raising and moving capital, these are industries where most of the cost is grunt work and compliance, where the reason a job needs ten people is that it needs ten people to push paper carefully, not to think ten people’s worth of thoughts. That’s exactly the work agents are about to absorb. The thesis I’m building toward is that the first great agent-native companies won’t be consumer apps. They’ll be the small, sharp teams that walked into a paperwork-heavy industry, automated the judgment-light layer entirely, and kept only the humans whose judgment was actually load-bearing.

I’d rather build that company than write think-pieces about it. So I’m building it, small on purpose, with the leverage pointed at the boring, valuable work that the rest of the world still thinks requires a crowd. The company is getting smaller. The ambition is getting much, much larger. Those two things only sound like a contradiction if you still believe headcount was ever the point.