Good luck getting rid of humans
The news that management is dead may be hugely exaggerated
Every time I open my feed, it’s the exact same mid-wit VC take parroted over and over: “The Agentic Era is here. Teams of 3 people will build $500M companies. Management is dead.”
It’s a beautiful fantasy, but it’s mostly pushed by people who have never actually run a business with a real payroll, angry customers, complex logistics, mildly dissatisfied employees and the likes.
Is management really dead?
The first thing they miss is that if a 3-person team can build a $500M company, they don’t need venture capital! I’m not sad about it, but I would point out that by hyping up this future, VCs are literally cheering for a world where their own business model is obsolete.
If you don’t need capital to scale, you don’t need investors.
I have spent the last 20 years in the trenches, scaling consulting organizations across 4 continents, looked at hundreds of P&Ls, and dealt with real corporate sagas. And I’m telling you: this idea that managers won’t matter in an AI world is completely laughable.
More efficiency = more things that don’t work
Silicon Valley purists love talking about “Agent Orchestration Layers”, basically AI managing AI. They genuinely believe Team A’s autonomous agent will seamlessly negotiate timelines and APIs with Team B’s agent, completely eliminating the human middleman.
Sure, for basic, mind-numbing technical coordination, they will be possibly right. The “just checking in on your status” manager is dead. Good riddance.
But they are missing a fundamental rule of economics called Jevons’ Paradox (I discussed it in this LinkedIn post). Historically, when you make a resource more efficient and cheaper, people don’t use less of it; consumption absolutely explodes.
Because AI makes output 10x cheaper and faster, you are not going to see smaller companies but actually you are going to see a massive explosion of micro-initiatives within companies. Instead of 2 massive, bloated departments, a company is going to have 50 hyper-lean, specialized human-agent squads.
And that is where the real nightmare begins.
More teams mean massive fragmentation. Agents can sync data endpoints all day long, but they can’t handle the stuff that actually moves markets, stuff like:
Political alignment: Who gets credit when two different AI squads build overlapping features?
Resource competition: Deciding which squad gets the real-world marketing budget when both have “perfect” AI-driven projections;
Brand preservation: Stopping a rogue autonomous squad from launching a product that accidentally cannibalizes your core business.
When you have 50 micro-teams all sprinting at 200 km/h in different directions, competing for the same internal capital and brand equity, who steers the ship?
Humans.
The next tech-optimist talking point is: “Well, baseline LLMs are a utility, but the companies with proprietary data flywheels will be unreachable.”
But… But… Everyone has access to data, or they can synthesize it, or they can buy it.
The real moat is not the data!
The moat is the organizational design that allows your people to direct the machine better than the competition.
Because it’s becoming increasingly easy to build stuff (almost free), the hardest part of business will be editing, and that differentiator will default right back to human intuition and creativity. You categorically cannot take humans out of the equation and expect unique, market-defining innovation.
If everyone uses the same perfect AI inputs, everyone gets the same average AI outputs. To get an edge, you need human weirdness, gut intuition, and the so much popularized idea of taste.
Where do we go from here?
We are entering a hybrid period that’s going to take a solid 5 to 10 years to fine-tune. Managers aren’t going away, but their risk profile is skyrocketing.
Think about the leverage:
If a human employee messes up, it usually takes a few days or weeks to notice. The damage is localized.
If an autonomous agent messes up, it cascades across your entire global enterprise in 4 seconds. It can hallucinate a pricing error, update your global ERP, email 10,000 customers, wipe out your quarterly margin before you have finished your morning espresso.
The room for error is exactly zero.
High leverage means astronomical risk that needs better adult supervision.
The headcount of traditional middle management will probably shrink (though, let’s be real… remember how “Agile” promised lean teams and just gave us 3x as many meetings? Corporate bureaucracy always finds a way to survive.)
But the managers who remain? Their importance will be off the charts.
They will be managing a high-velocity, algorithmic ecosystem fused with human psychology to become Editor-in-Chief, Risk Officer, Political Diplomat all rolled into one.
I don’t buy the hype that tech eliminates the need for leadership.
When the fog of the AI war settles, great management is likely to come out of it as the ultimate unfair advantage.
As usual, if you enjoy reading Consulting Intel, please do me a favor: spread the word and share this post.
👋
👀 Links of interest
A few corners of the internet you may find interesting:
This is a Note I posted recently… A quick taster of something cool to come soon 👀
Have you looked into the Leaders Toolkit? It is a deck of 52 tools, frameworks and mental models to make you a better leader (use code CONSULTANT10 for 10% off);
The Consulting Intel private Discord group with 250+ global members is where consultants meet to discuss and support each other (it’s free).
Very interesting! I know it will take many years to unfold, but I am quite curious about where we will land with the org structures of the future. They have essentially looked the same for so many decades it will be so interesting to see how different it might truly become.