Exocapitalism Part 3 - Lift describes the direction of travel. Coase explains the limit.

As per the previous posts, the first idea comes from Marek Poliks and Roberto Alonso Trillo’s Exocapitalism; the second from Ronald Coase’s 1937 essay on the nature of the firm. Coase’s insight was deceptively simple: firms exist because using the market is not free. Finding counterparties, negotiating terms, monitoring performance and dealing with failure all carry transaction costs. Work moves outside the firm only when market coordination costs less than managing it internally.

Every generation of technology has shifted that boundary. Standardised contracts, telecommunications, enterprise software, cloud platforms and digital marketplaces have progressively reduced the cost of coordinating beyond the firm. Agentic AI pushes the boundary again by potentially searching wider supplier markets, checking outputs and coordinating hand-offs continuously.

This is where the fox and the hedgehog become more than a dinner-party metaphor (thank you Mac Walker). They become an operating-model question.

Large organisations became foxes, knowing many things, because complex systems could not reliably be assembled from specialists in real time. A utility integrated engineering, operations, maintenance, safety, billing and service because breadth was often the least costly way to hold reliability and accountability together.

If agents genuinely reduce coordination costs, more organisations could afford to become hedgehogs: know one important thing, do it exceptionally well, and buy the rest from a market that can be reconfigured as conditions change.

That is a possibility, not a law.

Search and monitoring are only part of the cost. Liability, tacit knowledge, trust, cyber risk, physical safety and resilience do not disappear when execution is fragmented. In some cases, verification and dispute costs may actually increase.

There is another complication that deserves a board’s attention. The orchestration layer becomes a hedgehog of its own, with its one big thing being the coordination of everyone else: deciding who participates, how performance is measured, where work is routed and when payment is released.

Disaggregate production far enough and you may concentrate power at the point of coordination.

So yes, expect thinner firms in some domains. But test the proposition properly. Don’t ask whether an agent can complete the hand-off. Ask whether the system still works when a supplier fails, an instruction is ambiguous, the data is wrong, or someone has to carry responsibility for the outcome.

Coase’s boundary moves when the full cost of coordination moves.

It doesn’t move simply because the interface looks effortless.

#OperatingModels #AIGovernance #AusBiz