The Drucker Collapse

The Drucker Collapse, three mechanisms shown as a before / after grid. Row 1: pyramid with ember middle layer (knowledge-worker era) becomes pyramid with faded middle plus front-line-to-executive arrows (AI era). Row 2: solid firm-boundary circle with coordination activities ember becomes porous shifted boundary. Row 3: Person | Institution boxes with solid line become Person | AI system | Institution with dashed line.

​

Faint signals collapse into total signal. Peter Drucker’s middle layer existed because no senior leader could listen to everyone. Generative AI listens to everyone, in parallel, at machine speed, so the information rationale for that layer evaporates. The discipline shifts from gathering faint signals through bright sources to filtering total signal for what matters. What remains is a governance role, smaller and sharper than the work it replaced.

​

Transaction costs fall, and the firm boundary moves. Ronald Coase’s calculation of where to draw the firm’s edge depended on coordination being expensive. Generative AI makes coordination cheaper by an order of magnitude, so activities the firm did internally because external coordination was too costly become contestable again. The leader’s question changes from how do we run internal coordination better to which of our activities exist because of coordination cost, and which exist because of something else.

​

The Carroll line, drawn by you or by default. The Carroll line is the boundary between what the employee brings into the work and what the firm provides around it, the contract-of-employment question that generative AI has made urgent. When an AI system sits between the person and the work, two questions go live: what did the person bring, that they take with them when they leave? and what did the firm provide, that stays behind? The moment a senior hire walks, the precedent for who owns the prompts, trained patterns and curated context is set by whatever happens first. Most contracts do not yet say. Leaders who write the policy in advance keep the choice.

​

Once the three mechanisms are named, the leader’s job becomes legible: read which mechanism has shifted where, and apply the old or new reasoning to each part on its own terms. The Collapse is uneven; reading it as a uniform timeline moves you too early in the wrong places and too late in the right ones.

​

Genesis

Three sources stack into one frame. Peter Drucker, across Concept of the Corporation (1946) and Post-Capitalist Society (1993), argued that middle management exists to aggregate information senior leaders cannot gather alone. Ronald Coase (1937), The Nature of the Firm, supplied the boundary calculation: the firm exists where internal coordination is cheaper than market coordination. AI engineer Harper Carroll drew the third line: when an employee leaves, they lose access to internal systems but not the knowledge and judgment they have developed. Don Tapscott cited Carroll on the HBR IdeaCast With Rise of Agents, We Are Entering the World of Identic AI (February 2026) and extended it in Tapscott and Bradley’s You to the Power of Two (2026): the personal AI agent travels with the employee, the institutional knowledge stays with the firm. Each frame was right when written; each has been undercut, not refuted, by generative AI.

​

The moves

Re-specify the management role before resizing it. Name what fraction of each role is information aggregation, governance, and execution. Until that breakdown exists, any spreadsheet number is precise about the wrong thing. Map your activities on the Coase axis: coordination-cost activities are exposed; judgement-and-trust activities are durable. Draw the Carroll line, in writing, before you have to, in the role categories most exposed to AI augmentation. The first version will be wrong; the point is to have a position the firm can defend before the first dispute writes one for you.

​

How ChangeSchool applies it with executives

We run senior cohorts through a Three-Mechanism Audit: each leader brings one current restructuring proposal, names which mechanism is driving it, and re-specifies the analysis in the new vocabulary. The exercise reliably surfaces proposals built on the old reasoning’s clean numbers.

​

The discipline

A monthly governance audit on one role: tag activities as aggregation, governance, or execution, and track the ratio. A standing line in every restructuring brief naming which mechanism is driving the change. A quarterly conversation with HR and legal on the Carroll line, so the question stays live where the answer will eventually be written down.

​

​

​

​

‘The information rationale for the management layer has gone; the governance rationale has intensified; the leader’s job is to tell the difference.’ 

 

 

Viren Lall, Managing Director,

ChangeSchool LDN (2026).

virenlall.com/drucker-collapse

​

​

​

​

​

​