Your next manager may have three employees and 19 agents. Who's accountable?

TEXT START: Over the past several months, we have been interviewing CIOs and other senior executives about how AI is changing management.

The Dissection

The article is performing organizational anesthesia. It accurately observes that AI is decomposing management into information processing, workflow orchestration, agent supervision, access control, human development and outcome ownership. It then frames this decomposition as a redesign opportunity rather than as a mechanism for stripping labor of economic necessity.

Its central move is linguistic: replace “fewer managers and fewer workers” with “the management bundle is being reassembled.” The examples show that AI can absorb reporting, memory, monitoring, coordination and follow-through. They also show that the remaining human roles become more concentrated around exceptions, decisions, accountability and control. That is not evidence of broad labor preservation. It is evidence of rising leverage per surviving human.

The article’s real subject is not whether management survives. It is how institutions will redistribute authority after AI makes large portions of managerial labor unnecessary. Its proposed answer is task decomposition, governance frameworks and explicit ownership. That is useful for executives. It is not an adequate account of the social consequences.

The Core Fallacy

The article confuses the survival of managerial functions with the survival of managerial employment.

Under the Discontinuity Thesis, AI does not need to eliminate every management task. It only needs to perform enough coordination, analysis, monitoring and execution at lower cost and higher scale to destroy the mass employment circuit. A manager supervising three humans and 19 agents may be indispensable to the workflow while being economically valuable precisely because one person now controls the output of what previously required a much larger human organization.

“Management is expanding” is therefore a misleading description. The scope of responsibility expands while the number of economically necessary humans can contract. The surviving manager becomes a control node over a machine labor system, not proof that human labor remains broadly required.

The article also treats accountability as if assigning a human owner solves the underlying problem. It does not. A human signature can locate liability, but it cannot restore displaced productive participation, distribute ownership, or prevent organizations from concentrating output and decision power in the hands of AI-capital owners. Accountability is a governance requirement. It is not a jobs program.

Hidden Assumptions

• Organizations will use freed capacity for coaching, mentoring and service improvement rather than eliminating positions, increasing workloads or widening spans of control.
• Human oversight will remain economically necessary instead of becoming a thin legal shell around increasingly autonomous systems.
• “Human in the loop” will mean substantive judgment rather than ceremonial approval and liability transfer.
• Managers and individual contributors will be allowed to retain meaningful authority over agents rather than merely operate systems owned and controlled by senior capital holders.
• New responsibilities around agents will generate enough durable jobs to offset the human work removed from reporting, coordination, analysis and execution.
• Better outcome measurement will preserve human roles rather than expose how few people are required to produce the same revenue, margin or throughput.
• Organizations can maintain stable human-only economic domains despite competitive pressure to automate them.
• The redesign will be deliberate and humane, rather than governed by cost minimization, consolidation and shareholder advantage.
• Digital workers can be treated as subordinates without confronting the fact that they are capital assets with no wages, consumption needs or independent claim on the surplus they produce.
• Explicit ownership of outcomes is equivalent to a fair distribution of economic power. It is not.

These assumptions smuggle in P2—the belief that institutions can preserve meaningful human economic domains at scale—while ignoring P3: productive participation collapses when the majority are no longer necessary to produce the output society rewards.

Social Function

Primary classification: partial truth, transition management and ideological anesthetic.

The partial truth is real: management will be unbundled; information-heavy layers will be exposed; agent supervision and workflow governance will become genuine activities; accountability must be assigned explicitly.

The transition-management function is to give CIOs and CHROs a vocabulary for reorganizing labor without using the more destabilizing vocabulary of mass redundancy, ownership concentration or social exclusion. “Reassemble the bundle” is an executive-friendly way to describe a workforce being compressed around a smaller number of control points.

The ideological anesthetic lies in presenting the winners’ problem—how to redesign roles and govern digital labor—as if it were the whole problem. The article never seriously asks who owns the agents, who captures the surplus, what happens to workers whose tasks disappear, or whether the newly created high-leverage roles can absorb the displaced population. It turns a civilizational employment break into an org-chart exercise.

The accountability discussion is especially revealing. The human is retained as the accountable party because institutions need a responsible body to blame, regulate and sue. That does not mean the human retains productive power. The worker may become the liability-bearing wrapper around capital-controlled automation.

The Verdict

This is a competent corporate transition memo disguised as a neutral analysis of management. It correctly identifies the machinery of managerial decomposition and incorrectly implies that decomposition is primarily a redesign challenge.

The harsher reality is that AI is converting managers from labor coordinators into high-leverage supervisors of capital, while eliminating the lower layers that once made coordination a mass human activity. Three employees and 19 agents is not a wider version of the old workplace. It is a small human control surface attached to a much larger nonhuman production system.

The article sees the new control architecture. It refuses to follow that architecture to its terminal conclusion: management may survive as a function, accountability may survive as a legal demand, and human economic necessity may still die.