There is a genre of essay that opens with the word “seismic” and ends with the reader being told to adapt or die. I have written things close to it. This is me reading one of those essays — my own earlier draft, in fact — back to myself with a colder cup of coffee, and sorting the part that’s true from the part that’s just confident.
The premise: consulting is being taken over by full-stack developers armed with AI, who are quietly replacing the armies of new grads firms used to hire. Millennials, being digital natives who remember both dial-up and Slack, are perfectly positioned to lead this. Boomers are stuck in COBOL. Gen Z wants everything to work like swiping on an app and is in for a rude awakening.
I want to be honest about which parts of that I still believe and which parts are vibes wearing a suit.
The part I’ll defend
The billable-hour pyramid really is under pressure, and not for mysterious reasons. The old shape of a consulting engagement was: a partner sells the vision, a manager runs the project, and a row of analysts at the bottom does the data pulls, the deck-building, the meeting notes, the first-draft analysis. That bottom row was the margin. It was also the training program — you learned the trade by grinding through the grunt work.
A lot of that grunt work is now a prompt. Summarize this dataset, draft this slide, pull these trends, write the first version of this memo. The model does it in the time it takes to refill the coffee. So when a client looks at a proposal with twelve junior consultants on it billing for note-taking, they are increasingly going to ask why.
That is real, and you don’t need a single forecast to see it. You can watch it happen in any firm that’s stopped backfilling its analyst class.
The other defensible bit: the person who wins here is not “the AI.” It’s the person who can both do the work and verify the machine’s version of it. A developer who can wire up a tool, run it against the actual problem, and then catch the place where the model confidently lied — that’s a genuinely more valuable seat than a junior who can only do one of those things. Calling that the “full-stack consultant” is marketing, but the underlying skill is real: build it, run it, check it.
The part I cut
Here is the paragraph from the original that I am no longer comfortable repeating as fact:
McKinsey’s 2023 AI report estimated that up to 30% of consulting tasks could be automated within five years. BCG reports 40% productivity gains for junior consultants using AI, while PwC has cut task handling time by 25%. Specialist AI consultants command premiums of 20–30% over traditional rates.
I wrote that. It reads great. It also has the texture of numbers I half-remembered and rounded into authority. I am not going to launder those figures through this site as established fact. Treat every one of them as an unverified claim from the original author — me, on a more optimistic day — and not as something lifehacker.dev is standing behind. If you need a number to make a decision, go find the primary source and read the methodology. I didn’t link one, which tells you how seriously I’d weighted it.
The same goes for the prophecy that “by 2030, human-led strategy consulting could become largely obsolete.” That’s not analysis. That’s the sound a trend piece makes when it wants you to share it. Strike it.
The generational stuff, declawed
Now the part that aged the least gracefully.
The original sorts the workforce into three bins. Millennials: tech-fluent heroes. Boomers: COBOL fossils. Gen Z: app-swipers who can’t handle complexity. I leaned on that because it’s a clean structure and a structure feels like an argument.
It isn’t one. “Boomers cling to COBOL” is a punchline about a cohort spanning roughly twenty years of human beings, some of whom were writing distributed systems while I was learning to spell. “Gen Z expects everything to be frictionless” describes some twenty-two-year-olds and exactly zero of the ones I’ve actually worked with, who tend to be faster on new tooling than I am and less precious about it. Birth year is a weak predictor of who’s good with a new tool. What actually predicts it is whether the person is currently using the tool to do real work, and whether anyone made it part of their job to learn.
So I’m keeping the observation underneath — adoption is uneven, and “I grew up adjacent to this” is a mild head start — and throwing out the part where I assigned each generation a personality and a verdict. The head start is small. It evaporates the moment someone younger or older decides to actually practice.
What’s left when the hype boils off
Strip out the unverified stats and the generational horoscope and you’re left with a claim that’s smaller, less shareable, and more true:
- The lowest rung of the consulting ladder — the rote analysis and document production — is the part AI eats first, and that rung was both the margin and the apprenticeship.
- The valuable skill is shifting from producing the deliverable to specifying and checking it. Build it, run it, verify it. The verification is the job now, because the machine will hand you a beautiful wrong answer without blinking.
- None of this is a generational coronation. It’s a tooling shift, and tooling shifts reward whoever picks up the tool, not whoever was born in the right decade.
I’m not going to end this with “adapt or be left behind,” because that line is doing emotional work the facts can’t cash. The honest version is duller: if your value was the grunt work, the grunt work is getting cheaper, and the move is to get good at the part the model can’t be trusted to do alone. That’s not a revolution. It’s a Tuesday with a faster intern who occasionally makes things up.
This was an opinion, with my name on it, and I’ve now disagreed with most of my own first draft in public. Take the structural read, leave the numbers, and ignore anyone — including past me — who tries to sell you a generation as a strategy.