As an AI practitioner, I absolutely believe in and have architected AI systems like ACES that optimize workflows. However, I am in fundamental disagreement with the "Middle Manager Cull" taking hold in the tech world. This is not a question of being anti-technology; it is a question of structural reality and human performance.

If we imagine a 1,000-person company flattened to just three layers: a CEO, "Org Leads," and Builders, the structural strain becomes a matter of simple arithmetic. To scale this cleanly, the equation is:

Solving for x gives an answer of roughly 31. In this model, the CEO has 31 direct reports, and every lead is responsible for 31 people. While this looks streamlined on a slide, it creates a breaking point in practice:

• Executive Overload: A span of 31 forces a CEO to process raw noise rather than distilled decision points.
• The Triage Trap: A lead with 31 reports cannot coach or develop talent. They are merely firefighting, and mentorship fades into transactional feedback.

The ideology seen in Meta’s reframing of managers or Block’s vision of a player-coach model assumes that AI can replace the human coordination layer. This reflects a misunderstanding of how organizations actually function.

The "flattening" narrative is a delusion for several reasons:

• Decision-making is a judgment problem. It occurs inside social, cultural, and political systems that AI cannot navigate.
• AI cannot "read the room." It lacks the situational awareness to detect informal power, fear, or hesitation.
• Execution moves at the speed of trust. Trust and alignment are human requirements that do not move on raw information alone.
• Managers are contextual buffers. They absorb ambiguity from above and convert noise into clarity for their teams.
• Interpretation is not waste. Removing managers removes the very layer that builds the narrative making action possible.

AI will change management by automating reporting and reducing administrative drag. But it does not carry the emotional weight of difficult judgment. It does not know when a technically correct answer will fail because the organization is not ready to hear it.

The companies that succeed will not be those that strip away middle management as overhead. They will be the ones that use AI to make managers better: less buried in administration and more focused on clarity, coaching, and execution. Leadership work does not disappear just because task execution becomes more efficient.