When agents do the work your juniors learned on, how do you still build senior talent?

You noticed it in the last hiring round, though you might not have said it out loud. You did not open the two graduate roles you would have a couple of years ago. The agents handle the first-draft work now: the reconciliations, the early research, the tickets a new starter used to cut their teeth on. It feels efficient. And then a slower thought arrives, somewhere between quarters: if nobody is doing the work juniors used to learn on, where do your future seniors actually come from?

Here is the shape of it. The bottom rung of the ladder is being automated first, and that changes the whole ladder. PwC's 2026 AI predictions, drawn from a survey of 300 senior executives, find 79% already adopting AI agents (software that plans and carries out multi-step tasks on its own, rather than answering one prompt at a time), with two-thirds of those adopters reporting real productivity gains. PwC's own forecast is blunt: the workforce is losing its pyramid shape. The work is not disappearing. It is moving. The leadership task is to redesign how people grow when the first rung is gone.

Is AI really taking entry-level work, or does it just feel that way?

It is measurable, not a feeling. Stanford's Digital Economy Lab, working from ADP payroll records covering millions of workers, found a 13% relative decline in employment for people aged 22 to 25 in the most AI-exposed occupations, such as software development and customer service, since generative AI arrived at scale. The declines cluster precisely where AI automates the task rather than assisting the person. Economy-wide employment kept growing; the young end of the ladder did not.

This is the part worth sitting with. The tasks we hand to agents first, the structured, repeatable, checkable ones, are exactly the tasks a junior used to learn judgement on. Automate the apprenticeship and you keep the output while quietly losing the mechanism that made your seniors. It is the same pattern behind the shift in the engine of growth itself: the work that can be industrialised is going first, and the human value moves up a level.

What does the new shape of the workforce look like?

Not a pyramid. PwC predicts two new shapes, depending on the work. In knowledge work, an hourglass: strong at the junior entry point and at senior expertise, but thin through the middle, where agents absorb the coordinating and drafting that mid-level roles used to own. In frontline work, a diamond: agents take the entry-level tasks, and more mid-level people are needed to orchestrate them. Two-thirds of executives, 67%, expect agents to drastically reshape existing roles within a year.

Notice the tension, because it is the whole point. PwC's hourglass assumes firms use agents to make juniors more capable. The Stanford data shows many firms doing the opposite: using agents to remove juniors altogether. Same technology, opposite outcomes. Whether your organisation ends up with a thriving junior tier or a hollow one is not a fact you receive from the market. It is a choice you make in how you deploy the agents. The leaders who win the next decade are the ones who upgrade how their people work, rather than simply subtracting the people.

How do we keep building senior talent when agents do the junior work?

You build it on purpose, because it will not happen by accident any more. The old model, where people absorbed judgement slowly by doing years of starter work, is being dismantled task by task. So replace the mechanism deliberately, in this order.

1. Decide the shape on purpose. Your org chart in three years is a decision you make now, not a trend you absorb. Name the tiers you intend to keep strong before efficiency quietly removes one.
2. Use agents to accelerate juniors, not delete them. Point the agents at the grunt work so a junior reaches real judgement in months rather than years, rather than so the junior never arrives.
3. Move apprenticeship up the stack. The starter tasks that used to teach judgement are automated, so teach judgement directly and earlier. Pair every junior with both an agent and a senior, and make reviewing the agent's work the new training ground.
4. Protect the middle deliberately. The hourglass thins exactly where future leaders are forged. Ring-fence the mid-level roles that build orchestration and decision-making, and treat them as pipeline, not overhead.
5. Redesign the process, not just the headcount. Decide where a human must add judgement in the workflow, design the loop the agents run inside, then staff to that shape rather than cutting to last year's.

What does this change for me as a leader this quarter?

It reframes the efficiency you are being sold. Cutting the graduate intake looks like a clean saving this year and reads as a hollowed pipeline in five. That trade sits with you, not with the platform team, because only you can weigh the cost that will not appear on this quarter's numbers. And the value is far from guaranteed elsewhere: only 12% of CEOs told PwC that AI delivered both cost savings and revenue growth in the past year, so nobody has earned the right to treat headcount as the easy lever.

There is a longer arc behind this. Phase One, the Age of Effort: work hard, get a little more, linear growth. Phase Two, the Age of Scale: build once, sell to many, exponential growth. Phase Three, the Age of Acceleration: output decoupled from human effort almost entirely, the phase agentic AI begins to deliver. Phase One was muscle, Phase Two was machine, Phase Three is mind, and mind is exactly what an apprenticeship used to build. If you still want senior judgement in ten years, the way people grow into it is the thing to upgrade now, while the choice is still yours to make.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.