AI Is Eating the Bottom of the Consulting Pyramid

McKinsey now runs roughly 20,000 AI agents. Eighteen months ago it ran almost none. That's not a productivity story. That's a structural one — and if you've spent 20 or 30 years building deep expertise in a field, it's the most important signal in the market right now.

Because the firms that have always sold expertise for a living are quietly rebuilding how they do it. And the way they're rebuilding tells you exactly where the value in knowledge work is moving.

Let me walk you through what's actually happening, why it matters far beyond consulting, and what the smart move looks like if you're a senior professional watching all of this from the outside.

The pyramid was always the business model

The Big Four — and McKinsey, Bain, BCG right alongside them — never really sold "consulting." They sold a structure.

Here's how it worked. A partner lands a client. The actual work — the research, the data pulls, the slide-building, the financial modeling, the late nights — gets pushed down to a base of junior analysts and associates. Those juniors are billed out at rates far above what they're paid. The spread between what the client pays for their hours and what the firm pays them in salary is the profit. Stack enough juniors under each partner, and you have leverage. The wider the base of the pyramid, the richer the top.

That's the whole game. It's why these firms hire enormous classes of new graduates every year and burn through most of them. The juniors aren't there because the work requires brilliance. They're there because the work requires volume, and volume billed at a markup is how partners get paid.

For 50 years, that model was untouchable. The grunt work had to be done by someone, and a human was the only thing that could do it.

AI just removed the bottom three rows

That assumption died quietly over the last 18 months.

The research, the first-draft analysis, the data synthesis, the deck-building — the exact work that filled a junior analyst's day — is now what an AI agent does in minutes. So the firms are doing the math everyone in their position eventually does.

McKinsey's agent fleet didn't grow 500% in a year and a half by accident. Deloitte is scrapping its traditional job titles, a change that took effect June 1 — a tell that the old analyst-to-partner ladder no longer describes how work moves through the firm. EY has committed $1.4 billion to AI. KPMG, $2 billion. Accenture has put $3 billion behind its data and AI practice.

When a company that profits from junior labor spends billions to automate junior labor, read it for what it is. They are kicking the base out from under their own pyramid — on purpose — because the base is no longer where the margin is.

The value didn't vanish — it moved up

Here's the part most people miss, and it's the part that matters most to you.

When AI eats the junior work, the value doesn't disappear. It relocates. And it relocates upward — toward the things AI still can't do.

Think about what's actually left once the grunt work is automated. Knowing which question is the right one to ask in the first place. Reading a room. Knowing which client is telling you the truth and which one is managing you. Recognizing when the model's confident answer is confidently wrong — and having the scar tissue to know why. Deciding what to do when the data is ambiguous and someone has to own the call.

None of that is junior work. That's judgment. And judgment is the accumulated residue of 20 years of being in the room when things went sideways. A language model can generate a competent first draft of almost anything. It cannot generate 20 years of having been wrong and learned from it.

So the market isn't destroying the value of experience. It's stripping away everything around experience — the throughput, the volume, the busywork — and leaving the experience standing there, exposed and, frankly, repriced higher.

Watch who's leaving on their own

If you want proof of where the value is going, don't look at the layoffs. Look at who's walking out voluntarily.

The most interesting companies in professional services right now are the AI-native consulting shops — and a striking number of them are being founded by ex-Big Four partners. People who had the corner office. People who could have ridden it out. They left, because they did the same math the firms did and reached a sharper conclusion: if AI does the throughput, you don't need the pyramid underneath you to deliver real value. You need your judgment, a few good clients, and a stack of AI tools doing the work a team of analysts used to do.

They're not competing with the machine. They're sitting on top of it, pointing it.

And the technology companies see the same opening. OpenAI and Anthropic are openly moving toward the trillion-dollar market for management consulting and technical advice, with offerings that put strategy-grade help a prompt away. When the toolmakers themselves start eyeing your incumbents' core business, the incumbents' moat is already cracking.

This isn't just a consulting story

It would be a mistake to file this under "interesting news about consulting firms." Consulting is just the cleanest example because its business model is so nakedly built on the pyramid. But the same force is rolling through every field that runs on leverage and junior throughput.

Law firms run a pyramid. Accounting runs a pyramid. Marketing agencies, investment banks, engineering orgs, research departments — anywhere senior people are funded by an army of juniors doing volume work, the same repricing is coming. The base automates. The middle thins. And the value concentrates at the top, in the people whose contribution was never the volume in the first place.

You can already see the repricing in the data. Demand for fractional CFOs jumped 103% over the prior year, into a market north of $3 billion — companies renting senior financial judgment by the engagement instead of buying a full-time seat. That's the same pattern playing out one field over: the routine finance work gets automated or pushed down, and the judgment gets unbundled from the full-time job and sold on its own. Finance is simply a little further down the road than the rest. Your field is on the same road, whether or not the headlines have caught up to it yet.

That's the signal underneath the headlines. Every "AI layoff" you read about is the same market quietly answering the same question: now that AI does the throughput, what is a human actually for? The answer keeps coming back the same. Humans are for judgment, relationships, and ownership of the call. Everything else is getting automated or repriced toward zero.

What this means if you're a senior professional right now

If you're 50-something, deep in your field, and you've either been cut or you can feel the ground shifting under you — I want you to hear the reframe, because the story you're telling yourself is probably the wrong one.

You are not behind. You are not obsolete. You are sitting on the exact asset this entire shift is concentrating value into. The problem was never your experience. The problem is that for your whole career, your experience was bundled inside a W-2 job, delivered through a company's pyramid, and largely invisible to the wider market.

The shift that's threatening the junior analyst is the same shift that's repricing you upward. AI handles a huge share of an entry-level person's tasks and almost none of a seasoned operator's. The accumulated judgment you've built is the moat — but a moat only protects you if you put it somewhere the market can see it and pay for it.

That's the work. Not "learn to code." Not "reinvent yourself." Take the judgment you already have, get fluent enough with AI to point it instead of fearing it, and make that combination available to the market on your own terms.

How to read a layoff before you trust it

Not every cut means the same thing, and learning to tell them apart is its own edge — whether you're deciding where to send your energy or just trying to make sense of the noise.

When a company lays off and then talks specifically about the outcomes it's driving — the work it now does better, faster, or differently, the bets it's funding with the savings — that's a company with a real strategy. It made a hard call in service of a direction. Those companies can be worth approaching, because they know where they're going and they'll know where you fit.

When a company lays off and the explanation is all narrative and no substance — vague gestures at "AI transformation," dashboards showing usage is up, a story written for Wall Street rather than for the business — that's a company telling a story it can't yet back up. Watch what tends to happen next: a quiet round of rehiring a few months later, because they cut muscle they actually needed. Those are the firms to be wary of. If they didn't know who they could afford to lose, they don't know what they're building.

The tell is simple. Strategy layoffs talk about outcomes. Distress layoffs talk about activity. Learn to hear the difference and you'll know which side of this market to walk toward — and which to walk away from before it pulls you down with it.

The move: don't pick a lane, run both

Here's where people freeze, so let me say it plainly.

You don't have to quit anything today. You don't have to hang a shingle, print business cards, or become a salesperson overnight. The instinct to treat this as a terrifying all-or-nothing leap is exactly what keeps good people stuck.

The smart ones run two tracks at the same time. They keep the W-2 conversation alive — interviewing, networking, staying open to the right full-time role — and in parallel they start building an independent practice. A couple of advisory conversations. One small engagement. A point of view published where the right people can see it. Three companies that benefit from the same expertise instead of one company that owns all of it.

Some people call that independent work. Some call it advisory. Some call it fractional. The label genuinely doesn't matter. What matters is the structure of the move: stop letting a single employer own 100% of your expertise, and start packaging that expertise so the market can rent it — with AI doing the throughput a junior team used to handle, so you can deliver senior-level outcomes without a pyramid underneath you.

Run both tracks and you remove the fear. If the W-2 lands first, great — you walk in sharper, with proof you can operate independently. If the practice catches first, you've built an income stream that doesn't depend on any one company's headcount decisions. Either way, you stop being a passenger.

What an independent practice actually looks like

When I say "build a practice in parallel," I know what some of you picture: cold calls, a logo, a website, hustling for clients like a kid out of business school. Drop that image. It's wrong, and it's the thing keeping you from starting.

In practice, the first version is almost invisible. It's one former colleague who asks you to look at a problem and you say yes — for a fee this time, not a favor. It's a point of view you write down and put where the right people can see it, so the next conversation comes to you. It's one company that wants your judgment a few days a month, not five days a week. None of that requires you to quit, and none of it requires you to become someone you're not.

The AI part is what makes it work without a team. The research, the first drafts, the analysis, the deck — the work that used to require junior support — you now point an AI to do. That's how one seasoned person delivers what used to take a small group. You're not doing more grunt work in your sixties than you did in your thirties. You're doing none of it. You're doing the judgment, and the tools are doing the rest.

Start that way and the word "fractional" stops being scary. It's just the name for the thing you're already doing: your expertise, available to more than one buyer, on your terms.

Where to actually aim

The last piece is the one almost nobody gets right: which companies.

Because not every company is cutting. While the headlines scream layoffs, a whole different set of companies is doing the opposite — raising money and hiring as fast as they can. And here's the kicker: those companies are short exactly the kind of seasoned domain judgment you spent two decades building. They have capital and momentum. What they don't have is gray hair in the room.

Those are the companies to chase. Not the ones running layoffs to tell Wall Street an AI story — those are signals of distress, and you don't want to land somewhere only to be the next line item. You want the funded, growing, hiring side of the market, where your experience is the scarce ingredient.

The trouble is, that list isn't sitting on a job board. So I built it — and I keep it current.

Where to aim next

The funded, hiring companies — not the ones in distress — are where your next move is hiding. The funding index tells you who just raised and who's hiring right now. The guide maps the nine domains where your experience plugs in. Both are yours as a subscriber.