Judgment Doesn't Scale
Everyone’s telling the same story about AI and consulting right now. The big firms are thinning their lower ranks, the pyramid is shrinking, and the twenty-three-year-old with a spreadsheet and a framework is first out the door. If you run a company and you’ve wondered whether it’s still worth paying a firm to help with your AI push, some version of that story is rattling around your head too.
The disruption is real; it’s just not doing what the headlines think it is.
Start with what’s true, because plenty of it is. Accenture cut about 11,000 jobs in a single quarter last year, openly saying it was exiting people it couldn’t reskill around AI while it built its AI and data ranks up to around 77,000, nearly double where they stood two years earlier. McKinsey and the rest have been trimming and slowing hiring too. The entry-level analytical work, the research synthesis, the modeling, the deck production, is exactly the kind of thing these tools now chew through in a fraction of the time. Nobody’s wrong about that part.
Accenture, the firm leaning hardest into all of this, still expects its overall headcount to grow this year . Those bodies are getting swapped for different ones. The labor numbers are a distraction. What’s actually changing is what you were paying them for in the first place.
Consulting sold leverage. A partner you trusted sold the work, an army of smart, cheap junior level people would come in and do the work and the consulting firm marked up the difference. That gap was the product. The frameworks, the readiness assessments, the forty-slide decks, all of it was throughput you rented because producing it in-house was slow and expensive.
AI just made that throughput cheap for everybody. What a team of junior consultants used to bill three weeks for, one person can now draft in a couple of days with the right prompts. So the premium on leverage, the whole economic engine under the model, is quietly draining out.
What’s left is the part that was always the real value and never the line item on the invoice: judgment. Knowing which work is worth doing , what not to build, and where the model is confidently wrong enough to cost you six months if you believe it. I’ve watched plenty of clean, expensive analysis get produced perfectly and answer a question nobody should have asked. That’s the scarce thing now, and it’s uncomfortable for everyone in my line of work. Judgment doesn’t scale. You can’t stack it into a forty-person engagement, or train it into a junior level consultant over a summer, or mark it up. It sits with one experienced person who’s actually done the thing , and it won’t multiply no matter how you staff it.
None of this means the big firms lose. They still hand you things a solo operator can’t: genuine analytical breadth when a problem is genuinely bigger than one person can cover, a bench deep enough to survive someone quitting mid-project, and a name on the contract, so that if it all goes sideways, it wasn’t your call, it was theirs. That cover is worth real money and it isn’t going anywhere.
The shift runs underneath all of that. Big or small, everyone now earns on judgment instead of hours. A big firm that puts a genuinely wise person in the room is still worth every dollar. A solo consultant who’s really just renting you hours, running the same frameworks a model now runs for free, is selling the thing that got cheap, same as the pyramid.
Which changes the question you should ask when you hire help. The old questions were about capacity. How many people can you put on this, how fast can you deliver, whose logo goes on the letterhead. Those were the right questions when the work was the hard part. The work is the easy part now.
What you actually want to know is whose judgment you’re renting. When someone’s across the table pitching your AI strategy, the tell is simple: ask them where they’d tell you not to use it. Anyone selling capacity will keep selling. The person worth hiring will have a list, because they’ve watched this stuff fail, and they’ll walk you through the parts of your plan that are a bad idea before you’ve signed anything.
So AI is forcing the whole trade, me included, to finally answer a question we’ve mostly ducked: what was a consultant ever for? The work was only ever the wrapper. What you were paying for the whole time was the judgment, and now that the wrapper’s free, we’re all about to find out whose judgment was ever worth the price.