Box founder Aaron Levie said something recently that resonated across the industry: some tech CEOs have a form of psychosis when it comes to AI. They are so far removed from the actual work of implementing it that their decisions have detached from operational reality.
He is right, and the consequences are already visible.
The theory is seductive
The pitch from the top is consistent. AI will make small teams as productive as large ones. AI agents will replace repetitive work. Headcount can come down while output goes up. The org chart of 2027 will look nothing like the org chart of 2024.
ClickUp's CEO, Zeb Evans, put numbers on it: AI agents have created a hundred times more than the company's human workforce. ClickUp has deployed around 3,000 internal agents. The implication is that humans are being outperformed by software at scale.
Uber has cut staff while reporting that AI has reduced product launch times. GM is shedding IT workers and hiring AI specialists. Roughly 150,000 tech workers have been laid off this year.
The theory says this is a transition. The reality, for the people involved, is a firing.
The desire outpaces the capability
The problem is not that AI cannot do useful work. It can. The problem is that most organisations are not structured to absorb the change at the speed their executives are demanding.
Adopting AI is not like switching to a new software tool. It requires rethinking workflows, retraining teams, rebuilding quality assurance processes and accepting a period of reduced reliability while new systems stabilise.
Most companies are skipping that middle phase. They are cutting headcount based on the projected capability of AI tools, not the demonstrated capability. They are betting on the roadmap, not the product.
That is the psychosis Levie is describing. Not that AI is overhyped, but that the distance between the C-suite's expectations and the organisation's readiness is so large that decisions are being made in a fantasy version of the company rather than the real one.
The pivot that is not a pivot
The language of disruption makes this sound like a strategic pivot. It is not. A pivot implies a deliberate, controlled change of direction. What is happening in most organisations is a top-down mandate colliding with a workforce that has not been given the tools, training or time to execute it.
The executives read the research. They see the demos. They talk to other executives who are making the same cuts. The consensus reinforces itself, and the gap between the boardroom and the factory floor widens.
The sustainability question is real. AI tools are expensive. The costs fluctuate. The true long-term price of running thousands of agents at enterprise scale is not yet clear. Companies are making permanent staffing decisions based on temporary cost assumptions.
What happens when the bet fails
Some of these bets will pay off. Companies with strong technical leadership, realistic timelines and genuine process redesign will emerge leaner and more capable.
Others will find themselves understaffed, over-reliant on tools that do not perform as promised, and facing the awkward task of rehiring the people they let go, if those people are still available.
The AI transition is real. The productivity gains are real. But the speed at which companies are acting on theoretical capability rather than proven performance is creating a fragility that will become visible the moment the cycle turns.
The psychosis is not believing in AI. It is believing your organisation can transform at the speed of a keynote presentation. It cannot. No organisation can.
The gap between ambition and execution is where careers end, products break and companies discover that the future they were sold arrives on a very different schedule to the one they planned for.
