I’ve realized something uncomfortable about myself lately:

I’ve stopped mentoring juniors. Not because I don’t care—but because training AI is just… faster.

And I don’t think I’m alone.

The Part of the AI Story No One Talks About

This is the part of the AI story no one talks about. Everyone debates whether AI will “replace jobs.” However, the quieter shift is already happening: AI is replacing the mentorship that used to create those jobs in the first place.

We hired a junior employee a few months ago. Smart, eager, willing to learn. The kind of person who, five years ago, I would have spent hours mentoring—sitting in meetings together, walking through decks, reviewing their work line by line.

But I’ve noticed something.

The Time That Disappeared

I’m not spending that time anymore.

Every hour I spend giving feedback to a new hire, I find myself thinking: I could build this into a skill file for Claude and never explain it again.

The math keeps running in my head, even when I don’t want it to.

Training a person: repeat the same feedback weekly. Watch them make the same mistakes. Wait months for them to internalize patterns.

Training an AI: one feedback session. One skill file. Immediate execution. No drift. No emotional days.

Honestly, I don’t like this math. But I can’t unsee it.

The Silent Choice Every Manager Is Making

Every manager I talk to is quietly facing the same tension. They won’t say it out loud—it sounds cold. Nevertheless, the time they used to spend mentoring juniors is slowly migrating toward building AI workflows.

And this creates a problem no one is talking about.

Junior talent isn’t struggling because skills are harder to learn. If anything, AI made learning easier. You can teach yourself almost anything with YouTube and ChatGPT.

Instead, the struggle is different.

It’s that no one is investing in them anymore.

Why the Apprenticeship Model Is Breaking

The apprenticeship model—seniors spending years passing down intuition, correcting mistakes in real-time, building someone up through repetition—is quietly breaking. Not because seniors are cruel. Rather, it’s because the ROI equation changed.

When I spend 10 hours this week mentoring a new hire, I’m betting on a payoff 12 months from now. Maybe longer. Maybe never, if they leave.

In contrast, when I spend 2 hours building an AI workflow, I get the payoff today.

Most managers are making the same choice. They just won’t admit it.

Programmer: A Preview of What’s Coming

I keep watching what’s happening to software engineers. It feels like a preview.

For instance, junior developers are being squeezed out. Not because coding is harder—AI actually made it easier to write code. Instead, the squeeze is coming from a different direction.

Companies are realizing: Why hire someone who can learn to code, when I can hire an agent that already codes?

As a result, the entry-level pipeline is collapsing. Internships are disappearing. Meanwhile, the people still getting hired are architects—engineers who can design systems, not just write functions.

The same pattern is starting in sales. In product management. In marketing.

Specifically, the jobs that require “learning by doing under supervision” are exactly the jobs most vulnerable to this shift. Because the supervision is going away.

How the Career Ladder Broke

The old career ladder looked like this:

Junior (learn from seniors) → Mid-level (execute independently) → Senior (mentor others)

Each rung depended on the one above it. Seniors invested in juniors because they remembered being juniors. The system was self-reinforcing.

But AI broke the first rung.

If seniors stop investing in juniors—because AI is more efficient—then juniors never become mid-level. They get stuck. Or they leave. Or they’re never hired in the first place.

The ladder doesn’t disappear. Instead, the bottom rungs do.

What’s left is a jump: from “outsider” directly to “architect-level thinking.” No gradual climb. No hand-holding.

The New Hiring Reality

Companies aren’t hiring people who can be trained. Instead, they’re hiring people who already think like architects.

And if you’re a new professional, no one is coming to bridge that gap for you.

What Do You Do If You’re Early in Your Career?

I don’t have a neat answer. But I’ve been thinking about what I would tell myself if I were starting over today.

1) Stop waiting to be trained

The old playbook was: join a company, find a mentor, learn by watching and learning from others. That playbook assumed someone would invest in you.

Unfortunately, that assumption is increasingly wrong.

If you’re waiting for a senior to pull you aside and teach you the ropes, you might be waiting forever. After all, the people who used to do that are now building AI workflows instead.

The new default is self-driven. You teach yourself—using AI as your tutor, not your replacement.

2) Use AI to compress the learning curve, not to skip it

Here’s the trap I see junior people falling into: they use AI to do the work, instead of using AI to learn the work.

If you let Claude write your code, your emails, your analysis—you’ll ship faster. But you won’t build intuition.

The better move: use AI to explain why. Ask it to critique your thinking. Have it simulate a senior giving you feedback. Extract the mentorship you’re not getting from humans.

In short, AI can be your tutor. But only if you treat it as a learning partner, not an outsourcing machine.

3) Chase architecture problems, not execution tasks

The roles that are disappearing are the ones that say: “Do this task, then do the next one.”

In contrast, the roles that are growing are the ones that say: “Design the system. Decide what tasks should exist.”

If you’re early in your career, actively seek out architecture-level work—even if it’s not in your job description. Volunteer for projects where you define scope, not just deliver output. Push yourself into the uncomfortable zone where you’re making judgment calls, not following instructions.

Ultimately, the goal is to become someone who designs workflows—including workflows that involve AI—rather than someone who executes them.

4) Build in public. Create your own proof

The old career path gave you credentials through association. “I learned from so-and-so.” “I was trained at such-and-such company.”

If that path is closing, you need a new way to signal competence.

The answer is proof of work.

Ship side projects. Write about what you’re learning. Build a portfolio that shows your thinking, not just your resume.

When no one is recommending you, your work has to vouch for itself.

I want to be honest about something.

I’m not sure this is a better world.

The old apprenticeship model had inefficiencies. It was slow. It depended on the generosity of senior people. Not everyone had access to good mentors.

But it also worked. It built human capital. It created continuity. It gave people a way in.

What’s replacing it is faster and more efficient—but also colder. The burden of growth is shifting entirely onto the individual. Sink or swim. Figure it out yourself.

That’s not a moral judgment. It’s just what’s happening.

– Felix