Surface Area, Not Headcount

The instinct when you want to move faster is to add people. It is almost always wrong.

Every person you add is not just a pair of hands. They are a new set of edges. Every edge is a place where information has to cross from one head to another, and every crossing is a place where speed goes to die. Two people who each understand the whole problem will out-ship five people who each own a slice and have to coordinate. The five spend their day in the seams. The two spend their day on the work.

I think about this as surface area, not headcount. The question is not how many people are on the team. The question is how much surface each person can hold, and how few times a piece of work has to change hands before it is done.

Handoffs are the tax

Watch where a project actually slows down and you will rarely find someone stuck on hard work. You will find someone waiting. Waiting for a spec. Waiting for a review. Waiting for the one person who understands the other system to have a free hour. The work was not the bottleneck. The handoff was.

A handoff is a tax you pay in three currencies. You pay in time, because the receiver has to reload the context the sender already had. You pay in accuracy, because something is always lost in the translation. And you pay in ownership, because the moment a thing crosses a boundary, it becomes slightly less anyone’s problem. Add enough boundaries and you get an organization where everyone is busy and nothing ships.

What we optimize for instead

We keep teams small on purpose and we give people more surface than a bigger company would. A person here is expected to carry a problem from the fuzzy front all the way to the shipped end. That is uncomfortable if you are used to a narrow lane. It is also the whole point. When one person can hold the arc, the handoffs disappear, and when the handoffs disappear, the work moves at the speed of the work instead of the speed of the calendar.

This is not a claim that headcount never helps. It is a claim about sequence. You earn the right to add people by first removing the handoffs, so that the people you add expand what the team can do instead of expanding what the team has to coordinate. Add people to a high-friction system and you get a slower high-friction system. Add people to a clean one and you compound.

Why this scales with AI, not against it

This is the part that most people get backwards about agentic systems. The value of a governed agent is not that it replaces a person. It is that it removes a handoff. The work that used to wait in someone’s queue now moves the moment the conditions are met, under policy, with a human on the consequential calls. You are not adding a headcount. You are removing a seam.

That is why we build the way we build. Ten products run on one control plane, not ten teams reinventing the same spine. The leverage is in the shared surface, the parts every product needs that only get built once. Governance, identity, audit, the authorization model. Build that surface once and every product stands on it. Rebuild it ten times and you have ten teams coordinating instead of one platform compounding.

Small teams. Wide surface. Few handoffs. That is the shape of a group that ships. Everything else is coordination pretending to be progress.