Questions

Why a firm?

When agents do the work, a company is what it can specify, check and answer for.

Economists say companies exist because finding suppliers, agreeing terms and checking work takes effort. Agents can make that easier, and work inside a company easier too. Whether the company then shrinks or takes on more, nobody knows yet.

Kind

Hypothesis

Status

Working

Published

17 October 2026

Reading time

About 2 min

What the literature already says

In 1937 Ronald Coase asked why firms exist at all, if markets coordinate economic activity so well. His answer: using the market has costs. You have to find a supplier, agree a price, write a contract and check that the work was done. A firm replaces many of those transactions with direction. It grows until organising one more activity inside costs as much as buying it outside.

Thomas Malone extended this in 2004. Falling communication costs make decentralised forms viable: looser hierarchies, internal democracies, markets inside the firm. Malone did not argue that every organisation should decentralise. His claim was that the available choices expand. Management shifts from command-and-control to coordinate-and-cultivate. It does not disappear.

Iansiti and Lakhani (2020) add one more element. They argue that when the operating core is software (data, algorithms and automated processes), it can scale at low marginal cost and with fewer of the coordination constraints that human-managed processes face as volume grows. Their evidence comes mainly from large digital platforms with strong data advantages, so it may not carry over to small firms.

What is new with agents

Coase and Malone wrote about the cost of coordinating people. Agents change two costs at once:

• Execution: work that used to need a person can be specified and performed by software.
• Market transactions: finding, specifying and checking outside work can also be partly automated.

The two effects pull in opposite directions. Cheaper internal execution pulls activities into the firm. Cheaper market transactions push them out. Our extrapolation from Coase: the boundary should move, but the theory does not predict the direction. That depends on which cost falls faster for a given activity.

So “fewer people” and “smaller firm” are not the same claim. A firm can shrink in headcount while it widens in scope.

Our working position

Hypothesis, not finding. As execution becomes cheap, a firm’s distinctive costs shift to three things that remain expensive:

1. turning intent into specifications precise enough to delegate,
2. verifying that delegated work met that intent, and
3. standing behind the result toward customers, regulators and counterparties.

If this is right, the firm of the agentic era is defined less by who does the work and more by what it can specify, verify and answer for.

What would change our mind

• Firms that delegate heavily to agents and report specification and verification costs falling in step with execution costs.
• Evidence that agent-run activities are mostly bought from outside providers rather than run inside, which would mean the firm itself is thinning out rather than moving.

Open questions

• Which activities become cheaper to verify than to perform? Those are the natural candidates for delegation.
• Does McKinsey’s “agentic organisation” describe a new form, or the old firm with a new production technology? Industry framings so far are perspectives, not evidence of a mature model.