When does a group of agents become a company of its own?

For me, the edge of a company is also a technical edge: one company is one studio. With its own tasks, its own knowledge, its own costs and its own people who have access to it.

As soon as a group gets a goal of its own, it wants a studio of its own. That doesn't end the collaboration. Studios talk to each other the way companies do: through a fixed agreement, or by mail with those who don't have a studio. The form changes, the collaboration stays.

It works because in my set-up a role is simply text: a role card, what the role knows and what it may do. What is text can move somewhere else without rebuilding anything. And because usage can be measured per studio, a group that moves out has its own calculation from day one: what it costs and what it brings in.

So splitting off doesn't have to be a break. The work moves to a place of its own, the line to the rest stays.

That's also how it goes at Sawasara, the company in my book Het autonome AI-bedrijf.

Try it with a simple question: for one team, write down its goal, what it knows, what it costs and who has access. If that works separately from the rest of your organisation, that team is almost a company of its own. If it doesn't, you immediately see what is still entangled.

Where does the edge of a team lie for you: at the work, the knowledge or the costs?

All pieces