The coordinator was already an algorithm

Posted on: 8 January 2026

In 2025, Amazon laid off fourteen thousand corporate employees. That same year, leaked internal documents revealed plans to avoid hiring six hundred thousand warehouse workers by 2033. This is not a contradiction. It is the clearest photograph we have of a transformation that has inverted every assumption about what counts as "complex work."

For thirty years, the consensus held that automation would eat from the bottom up. Factory workers first, then clerks, then, perhaps eventually, managers. The opposite is happening. Management job postings sit forty-two per cent below their peak, with no recovery in sight. Forty-one per cent of companies reduced management layers in the past year. Microsoft publicly declared it wants "flatter structures with more engineers and fewer middle managers." This is not a weak signal. It is a structural earthquake that most people are misreading.

The paradox has a name: the Great Inversion. Artificial intelligence eliminates coordination roles faster than physical ones. It sounds absurd until you examine what middle managers actually did for the past fifteen years.

The answer, in most cases, is: translate. Translate executive objectives into tasks for execution. Translate execution problems into reports for the executive suite. Schedule, monitor, consolidate, redistribute. Activities that appeared to require human judgement but in reality followed repetitive patterns disguised as complexity. The middle manager had already become an algorithm before the real algorithm arrived.

When a tool does the same thing but costs less, there is no strategic debate. There is only a question of timing. And time has run out.

The temptation at this point is to find villains. Corporate greed. Executive short-termism. Technology hype. These explanations are comfortable and wrong. Not because these elements do not exist, but because they do not explain the mechanism. And without understanding the mechanism, any response will be ineffective.

The mechanism is this: for decades, organisations responded to complexity by adding layers. The more the business grew, the more coordinators it needed. The more coordinators grew, the more it needed coordinators of coordinators. This created what some call the "diamond-shaped company": a handful of decision-makers at the top, a handful of doers at the bottom, and a vast mass of translators in between. People paid six figures a year to ensure information flowed correctly between those who decide and those who do.

The problem is that information no longer needs human translators to flow. A system that analyses data in real time, generates reports, identifies anomalies and proposes corrections is not "helping" the middle manager. It is making the middle manager redundant. Not tomorrow. Now.

Anyone working inside a large organisation already knows this, even if they do not say it aloud. They know because they watch colleagues spend entire days in meetings that produce more meetings. They know because reports that once took three days now generate in three minutes. They know because the question "what exactly do you do?" has become increasingly difficult to answer without resorting to vague verbs like "coordinate," "oversee," "facilitate."

The instinctive reaction is to cling. Prove indispensability. Work longer hours. Be more visible. This is the wrong strategy, not because it is morally questionable, but because it does not work. You cannot compete on speed with a machine. You cannot compete on precision with a system that never sleeps, never gets distracted, never has an off day. If your value proposition is "I do what I did before, but better," you have already lost.

The right question is not "how do I remain relevant in my current role." The right question is "what can I do that a machine cannot, and that someone is willing to pay for."

The answer is not obvious, and anyone who tells you it is wants to sell you something. But there are clues. The roles that are growing, while everything else contracts, share common characteristics. They do not translate: they decide under conditions of radical uncertainty. They do not coordinate: they solve problems that have no standard procedures. They do not monitor: they intervene when the system breaks and no one knows why.

In other words: they do not execute algorithms. They design systems. They do not apply rules. They rewrite rules when the rules stop working.

It is the difference between someone who knows how to use software and someone who understands why that software exists, what problems it solves, what problems it creates, and what happens when it fails. It is the difference between following a process and seeing how processes interact with each other, where they jam, where they produce unintended effects.

This capacity cannot be learned in a training course. It is not a "skill" you add to your CV. It is a different way of looking at organisations: not as hierarchies to climb, but as systems to understand. Those who spent twenty years perfecting their ability to navigate hierarchy discover that hierarchy itself is dissolving. Those who invested in understanding how systems work, regardless of position held, find themselves suddenly at an advantage.

The point is not to learn how to "use artificial intelligence." Everyone will do that, and it will quickly become a baseline requirement, like knowing Excel in the nineties. The point is to understand what artificial intelligence cannot do, and position yourself there.

It cannot handle genuine ambiguity, the kind where no historical data exists to base predictions on. It cannot build trust with an angry client or a demoralised team. It cannot make decisions that require balancing conflicting values, where no "optimal" answer exists, only more or less acceptable trade-offs. It cannot see the problem no one is naming yet, the one that only emerges when you look at the system from a perspective no one else is using.

These are not marginal capabilities. They are the heart of what distinguishes an organisation that works from one that merely exists. And they are dramatically scarce, not because people lack them, but because the previous system did not reward them. It rewarded conformity, predictability, the ability to not create problems. Now that system is collapsing, and those with suppressed capabilities discover they hold an advantage they did not know they possessed.

The problem is that the transition will not be orderly. There will be no moment when the old system ends and the new one begins. There will be years of confusion, where companies lay off and hire simultaneously, where required competencies change faster than anyone can adapt, where someone with fifty years of experience finds themselves competing with someone who has thirty but understood earlier where the world was heading.

In this confusion, the most dangerous temptation is to wait for someone else to figure out what is happening and tell you what to do. It is the executor mentality: give me the rules and I will follow them. But the rules are being rewritten in real time, and whoever waits for instructions will always be one step behind.

The alternative is not the naive optimism of those who think "just reinvent yourself" is a plan. It is the clarity of those who look at the system for what it is, identify where it is going, and move before movement becomes mandatory. It is not a guarantee of success. It is simply the only strategy with a chance of working.

The coordinator was already an algorithm. The question is whether you were one too.