McKinsey secretly confirmed they're doomed.

Any developer knows that "25,000 subroutines" is not an indicator of productivity. Any designer knows that "2.5M PDFs generated" has nothing to do with impact. Any agile coach knows that "1.5M sticky notes" is an utterly meaningless objective.

And yet, here we are. McKinsey proudly claims to have "25,000 agents generating 2.5M charts and saving 1.5M hours."

For the uninformed: there is no meaningful way to count the number of AI agents.

You can build one bloated agent with a huge fragile context window trying to replicate the work of one human, but it will easily break due to what we call context rot. The agent gets confused. It has too much to do. The smarter AI chiefs among us know that it's best to break the work into smaller pieces: build mini-agents for subtasks and have an orchestrator on-the-loop dividing all the work. Suddenly you have ten "part-time" agents instead of one "full-time" agent, doing the same work much more reliably, and some of those agents will have nothing to do 99% of the time. But they will be very good at their specific task. We call this context engineering: finding the optimum for the division of labor, which needs a completely different approach for AI agents compared to humans.

In other words, the phrase "The goal is to have one AI agent per human" is utterly meaningless. It's the dumbest thing that anyone who claims to understand AI orchestration could post. The number of agents is whatever fiction serves your narrative.

I expect McKinsey knows this. They are well aware that their statistics are a big charade of productivity theater. But their audience is not people like us: smart chiefs building networked agentic organizations. Instead, McKinsey's customers are ignorant traditional executives who are very impressed by pointless bullshit statistics like, "40,000 humans + 25,000 agents = 65,000 employees."

In a way, you could say McKinsey is smart. After all, the universe has an unlimited supply of stupidity. The Total Addressable Market for ignorance is massive. The question is how soon that market will evaporate as managers of the future get more familiar with AI.

What I'm not impressed by is people uncritically amplifying McKinsey's claims as if we are witnessing the dawn of a new consulting paradigm. We're not. Instead, we're watching a legacy firm milk its last cohort of uninformed buyers before the future renders this whole performance obsolete.

McKinsey publicly admitted they're targeting fools with productivity porn, and the ignorant are lapping it up.

For more critical views of the industry, sign up for my newsletter.