Here’s a question I can’t stop thinking about, and I don’t think the sector has a good answer for it yet.
For generations, professional services have run on a simple, elegant machine. Bright graduates join, do the foundational work, the research, the first drafts, the data rooms, the bundles, and through thousands of hours of it, they slowly become the experienced people who run client relationships and, eventually, the firm. The grind wasn’t just billable. It was the apprenticeship.
Now AI is extremely good at exactly that foundational work. And firms, including some very well-known names, are openly reducing junior and support roles as a result. The short-term maths is obvious and tempting. But it raises a question that should worry every senior partner and HR director in the country: if a machine does the work that used to forge your future leaders, where do your future leaders come from?
I find this genuinely difficult, and I’d rather sit honestly in the difficulty than pretend there’s a neat fix. We can’t, and shouldn’t, recreate the old pyramid out of nostalgia. Asking young professionals to grind through work a machine can do better, purely so they “earn their stripes, ” is neither kind nor sustainable, and that generation can see straight through it.
But I also refuse to accept the bleaker version where we simply hire fewer juniors, work them around the automation, and quietly hollow out the pipeline. That’s a decision that feels efficient this year and catastrophic in fifteen.
There’s a third path, and it’s where my optimism sits. The job of an early-career professional doesn’t have to disappear: it has to be redesigned. Instead of doing the routine work, they learn to direct it, check it, challenge it and own the outcome. Reviewing AI output critically is a genuine professional skill, arguably a harder and more valuable one than producing the first draft yourself. It demands judgement earlier, not later. Done well, you could grow better professionals faster, not despite AI, but through it.
That redesign doesn’t happen by accident. It needs someone to rethink what a training contract or a graduate scheme actually develops, how supervision works when the “first draft” comes from a model, and how you keep young people learning the craft rather than just operating the tools. This is change management in its truest sense, not a software rollout, but a deliberate reshaping of how a profession brings its next generation through.
I care about this one personally. I’ve spent a lot of my career arguing that talent is everywhere but opportunity isn’t, and I worry that an unthinking response to AI could quietly close doors for exactly the people we should be opening them for. Get the redesign right, and AI could widen the path into these professions. Get it wrong, and we pull the ladder up behind us.
So I’d put the question back to any firm reading this: what is your graduate of 2027 actually learning to do? If the honest answer is “the work we’re about to automate, ” that’s not a reason to stop hiring them. It’s a reason to redesign the role around them, fast.
That redesign is squarely the kind of work we do with leadership teams. If your pipeline keeps you up at night, it’s a conversation worth having.
