The Death of the Manager

In 1941, James Burnham published The Managerial Revolution, arguing that the real rulers of the 20th century would not be entrepreneurs or workers, but managers. They did not own or invent, but administered. Bureaucracy itself would become power.

His thesis was so compelling that his friend George Orwell drew on it when writing 1984, with Burnham widely rumoured to be the inspiration for the novel's main character, O'Brien.

In any case, as with anyone trying to see the future, Burnham got plenty wrong. Yet on the unstoppable rise of managerialism, on this I think most readers would have to conclude that he was ever so slightly right. That over the past five decades we witnessed managerialism become the Leviathan.

Services have ballooned, corporations have calcified, and bureaucracy has seeped into every pore of work. We have built vast empires of compliance, reporting and process that would make Kafka blush. Whole industries whose output was not products or ideas but meetings, memos, and metrics.

Responsibility has diffused upwards to the high heavens, so far removed as so as nobody can be held to account. The machine of managerialism has churned on, even as productivity has flatlined the world over.

The truth of this was laid bare during COVID. Overnight, governments around the world admitted what David Graeber always called bullshit jobs: the vast majority of white-collar roles were declared "non-essential." Millions of workers stayed home, and much of the machinery of managerialist society simply stopped — with little visible impact. What kept the world running were the doers: delivery drivers, nurses, engineers, builders, farmers.

Then came the reckoning. People didn't return to work, inflation soared, production collapsed. The managerial class had presided over a system so bloated with administrative overhead that when the economy needed to flex, it couldn't. Suddenly, the principles of DOGE — the Department of Government Efficiency's ruthless pursuit of lean operations — began to metastasise across the Western world. What started as government reform became corporate gospel: cut the fat, automate the redundant, eliminate the unnecessary.

Remote work then only deepened the divide. Once people disappeared from the office, it became obvious how much of their labour was abstract, disembodied, and easy to sever. If a role can be managed without presence, it's not much of a mental leap to think how it can be managed without people at all.

I often write of how the cost of labour will collapse due to AI. But this will not happen overnight. The human-in-the-loop will be the state of play for some time.

That said, already, it is clear that AI is devouring the bottom rungs of the ladder. Junior roles built on tedium, repetition, and low-risk execution are the first to be absorbed. Drafting, scheduling, reporting, summarising: all the entry-level work that once justified vast pyramids of managers overseeing juniors is being done faster, cleaner, and cheaper by machines. Anthropic's CEO warns that up to half of entry-level office jobs could vanish within five years, and Stanford data already shows a 13% drop in employment for 22- to 25-year-olds in AI-exposed professions since 2022.

But it is the managers — the whitest of white-collar jobs — who are the clearest in line for automation. Because once the work of oversight, reporting, and coordination can be agentified, the very reason for the manager's existence collapses.

Large language models already touch 10% of the tasks of 80% of U.S. workers, and for nearly one in five, it's half the job. The question today is not what happens when AI changes work — it already has — but what happens when the human in the loop becomes the liability in the aim-to-outcome chain. This question is highly pertinent because the machine increasingly does not need a middleman.

The shift is subtle at first: fewer status meetings, fewer sign-offs, fewer hands touching a process. But each small hand-off removed is a rung kicked out of the ladder that once justified whole tiers of management. When coordination and reporting are executed instantly by software agents, the bottleneck isn’t the model it’s the human checkpoint slowing the flow.

In this light, the middle layer doesn’t just look redundant; it looks like friction. It’s a small shift on paper, but culturally it upends the very logic of how organisations justify themselves.

This isn’t some distant forecast but a restructuring already taking shape inside large organisations. As AI steadily absorbs coordination and reporting — because AI agents can now talk to one another and handle much of the workflow themselves — the old pyramid of work begins to flatten: fewer layers, faster loops between decision and delivery.

Harper Reed, CEO of 2389 Research, describes the emerging structure: companies will hire "junior employees who used A.I. to do what was once midlevel work, a handful of senior employees to oversee them and almost no middle-tier employees." Reed calls this the barbell economy — AI-augmented juniors at the bottom, strategic leaders at the top, and a hollow middle where managers once lived.

David Furlonger, a Gartner analyst who surveys CEOs, put it bluntly in a recent NYT piece: "C.E.O.s are implying in the data that we don't need as many of them [managers] as we did previously." He notes that AI can handle "anything that is administrative, spreadsheet-related, where there's an email trail, a document-management type activity" — the very tasks that justify managerial positions. What's left for managers is "more mentoring," but as Furlonger's data suggests, you need far fewer people for that.

However, it is worth noting that even this is a shoestring stopgap. The longer arc is toward organisations where autonomous agents handle almost everything — planning, execution, and adaptation — leaving only a thin layer of strategic oversight at the top. But I digress.

At Unilever, the new CEO declared he is "fed up with mediocrity," having already cut 18% of white-collar staff in just 18 months, and is now reviewing the top 200 leaders one by one — with a quarter expected to go. Earlier this year Microsoft layoffs specifically targeted "many middle managers", alongside software developers.

Amazon even went so far as to create a bureaucratic Bocche di Leone — a digital echo the lion-headed stone slots of medieval Venice where citizens posted anonymous denunciations of corrupt officials.

Back in La Serenissima, it was a mechanism to keep the bureaucracy lean and accountable. At Amazon, it serves the same end: letting workers slip notes about pointless processes and bureaucratic waste. The company says the mailbox has already led to 375 changes in process. Meanwhile AT&T's John Stankey issued memos demanding higher performance and fewer excuses.

What we’re seeing reflects a general mood: a broad revulsion against managerialism, with CEOs reaching for sharper sticks because the old playbook of committees and consensus no longer works. Mediocrity has become intolerable. Efficiency desperately needed. AI the obvious answer.

“It doesn’t matter whether a cat is black or white, as long as it catches mice.”
Deng Xiaoping

Across industries, CEOs have been spelling it out in blunt terms. Earlier this year, a trio of memos made headlines. Shopify's Tobias Lütke told staff that AI is now a non-optional skill: every employee, even executives, must prove they can use it or justify why AI can't. Duolingo's Luis von Ahn declared the company AI-first, arguing that urgency matters more than perfection — "we'd rather move with urgency and take occasional small hits on quality than move slowly and miss the moment." Fiverr's Micha Kaufman was even starker: "AI is coming for your job. It's coming for mine too."

These were internal memos, admissions that the old managerial order is cracking. The expectation is clear: AI isn't optional, it is the baseline.

Thanks to vibe coding and natural-language prompting, we now live in a time of software on demand. The brain-machine interface is melting. And the ability to build no longer belongs only to trained engineers. Anyone can spin up an app, automate a workflow, or launch a service in hours. What was once the guarded domain of technical specialists is becoming ambient capability.

The logic of the firm — hierarchies of managers coordinating scarce technical skill — starts to bend when anyone can make anything, and has little reason not to.

For the time being at least, herein lies individual opportunity. The same technologies that strip out layers of management theoretically make it possible for individuals to run not one career but several companies at once. With agentic AI handling operations, execution, and growth, a single person could spin up five or ten ventures in parallel. Each automated, each scaled, each operating on their behalf.

If the 20th century was the rise of the managers, the early Intelligence Age will witness their fall — replaced not by another bureaucracy but by networks of agentic systems, owned and directed by strategic, creative, and entrepreneurial individuals.

As I have written before, the stopgaps between what a machine can and cannot do shorten by the day. In the coming wave, no job will be untouched. Juniors go first because they are ‘easy’. Managers next because they are fat.

Yet for now the world still runs on middlemen. AI will wean us off them, and when it does, millions will be cut adrift with no clear place in the new order. History shows that upheaval always demands a bargain. The industrial age created factories, but it also created schools, welfare states, pensions. That was what made the disruption tolerable. So it must be again: disruption matched by delivery.

If the 20th century was the rise of the managers. The early Intelligence Age will witness their fall, replaced not by another bureaucracy but by networks of agentic systems, owned and directed by strategic, creative and entrepreneurial individuals.

The managerial revolution Burnham foresaw is now eating itself. What replaces it will have to be equally total, a new organising principle for how we live and work, in a world where what works wins, and humans may no longer work at all.