In the spring of 2026, Meta moved roughly seven thousand people into a new Applied AI division. Some of them had been managers. Then they weren’t. They became individual contributors, part of a larger bet that flatter was faster, that middle management was the soft tissue you cut when you needed to fund the machine.
A few months later, according to Business Insider, Meta started asking some of those same people if they’d like to be managers again. Opt-in. Quiet. A partial reversal of the year they’d spent praising lean, startup-style teams.
This is a confession.
For a decade, tech treated the manager as a tax on velocity. Fewer managers, smaller teams, more “makers.” Zuckerberg called it startup mode. The layoffs this year hit managers hardest. The theory was simple: AI would absorb the coordination work that used to require a human with a calendar and a title.
Then the work got hard. Training models at scale is not a solo sport. Someone has to define the role, set the A+ standard, run the one-on-one, give the small correction when the draft drifts, and decide who sits in the next seat. That someone is a manager. You can rename the job. You cannot delete the function.
At Esteemed, we teach a blunt definition: the purpose of management is to drive A+ results and retain the A+ team. Not to hold meetings. Not to sit between the work and the board. Great managers do three things. They know the individual. They communicate on performance. They ask for more. Their tools are boring on purpose: one-on-ones, specific feedback, goal setting, and delegation.
Those are not the enemy of AI. They are the operating system for it.
The premise of the Esteemed MBAi is that the same principles that earn the trust of a human team, clarity of role, consistency of feedback, discipline of delegation, are exactly how you calibrate an AI collaborator. You do not get leverage from a flatter org chart. You get leverage from a clearer one. Cut the managers, and you do not remove management. You push it onto people who were never trained for it, or onto an AI that was never given a Day 1 Charter.
Meta’s reversal is what happens when a company confuses headcount of managers with the practice of management. The first is a cost center. The second is how anything hard gets done, including AI.
Uber recently cut half its micro-teams. The pendulum is not swinging back to bureaucracy. It is swinging back to a quieter truth: someone has to own the people, the standards, and the handoff. If you are training models with thousands of humans in the loop, that someone is not optional.
So the question for any leader reading the Meta story is not “Should we have fewer managers?” It is the better question:
What has to happen for this team to be worth it, for the humans, and for the AI they lead?
Clarity of role. A standing meeting that isn’t theater. Feedback that lands before the iceberg. Goals you can measure. A successor on the bench. Management is not overhead. It is the work that makes the work compound.
Be Esteemed.