The limit case

Push the argument to the end and you get a company of one person and a thousand agents. It is worth doing the thought experiment properly, because it answers the question underneath all the others.

Strip the labour out and what remains is the list from the section above: proprietary data, the decision record, distribution and trust, and the tests that encode the firm's judgement. That is the honest inventory of a business once execution is rented by the token, and it is short. But the limit case exposes something the inventory hides. Those tests were built from the judgement of people who learned the work by doing it. With no people doing the entry-level work, there is no curriculum. With no curriculum, the judgement stops renewing. The evaluation set is frozen at the day the last practitioner left.

A one-person company with a thousand agents is a business running on stored judgement, and stored judgement depreciates.

That is why the three questions on the front door are one question. What the business owns is only worth something if the business can keep producing the judgement that owns it, and it can only keep producing that judgement if the agents doing the work are held to a standard a human still sets and can still check. Take any one away and the other two decay.

The intellectual-property version of the same point: the models are trained on everyone's data, so nothing you put into them differentiates you. What differentiates you is the state you keep on your own side of the boundary — the decisions, the corrections, the test set — and the people who keep it current. The societal version is the coordination failure from Room 1: every individual firm is rationally indifferent between the workforce that keeps learning and the one that doesn't, because the profits look the same for years. Nobody is going to fix that from outside. It gets fixed, if at all, inside the design of the work.

So the end state is not a company of one. It is a company that has decided, on purpose, which judgement it keeps producing in people, and built its agents around that decision rather than in place of it. That is a smaller firm than today's. It is not a firm of one.