What Will My Kids Do for Work?

With all the recent advances — Opus 4.6, Codex 5.3, and everything coming next — even the AI CEOs themselves are saying a great number of knowledge work jobs will be gone in a couple of years. I have three boys who'll be entering the workforce in the next three to ten years, and it makes me genuinely wonder what their prospects look like.

I don't have a clean answer. But I've been thinking about it a lot.

The Jobs I Did Are Already Disappearing

If the traditional desk jobs are gone — building PowerPoint slides, coordinating meetings, sending emails, pulling people and information together — that's a lot of what I was doing. At P&G as a Product Manager, I was setting product direction, but honestly a huge amount of my day was attending meetings, sending emails, and putting together status reports.

I was already able to leverage AI to do a lot of that for me. And it's only going to accelerate. The ability to take in a lot of information, summarize it, and coordinate activities — those skills are being automated right now. An AI that attends your meetings, writes follow-ups, tracks progress, sends updates. All things I did. All things that won't need a human much longer.

Even the "Human" Skills Are Getting Automated

What probably the AI wouldn't do anytime soon would be... actually, it's really hard to think of anything. When I look at everything I was doing, very nearly all of it could be done by AI.

Even product-level decisions. A lot of people have said that "taste" is something LLMs can't do, but I'm reading that people are starting to doubt that. The most recent models are genuinely great at product decisions, building roadmaps, prioritizing features. Those are all things I did.

Effectively, I think an agent could have replaced me in my last role. I really do.

Breadth vs. Depth in an AI World

I bring almost thirty years of experience across IT, so I have a big breadth of knowledge. But I'm honestly not super deep in any one area. And I wonder if that's actually a disadvantage now.

The advantage going forward might be the opposite of what it used to be: be super deep in one area and use AI to work broadly. AI is incredibly good at being wide — pulling in context from adjacent domains, synthesizing information, coordinating across disciplines. What it needs from a human is depth. Real expertise. The kind of judgment that comes from years of focused work in a specific domain.

That might be where engineering goes. Where product goes. People who are very deep, using AI to be very wide.

What I'd Tell My Kids

Companies are already reducing their engineering headcount. You can run complex platforms with a lot fewer engineers now, and it'll be even more so in the future.

So what's the play? I think the advantage right now is being super fluent with AI tools — knowing how to use them to build, to be productive, to ship. Maybe the best shot is building your own thing, because you can absolutely leverage AI to do that if you know how.

If I were in high school or college right now, here's what I'd focus on:

1. Be deeply passionate about a project or an idea. Not a job description — a problem you actually care about solving.
2. Get extremely good at using AI capabilities to build it. Not just prompting — the full stack of AI-augmented development, design, marketing, operations.

As the tedious coordination jobs go away, what's left is people who are deeply passionate about a product and can leverage AI tools to bring it to life. That's the job. That's the career.

The Honest Take

I don't know what's going to happen to the job market. Nobody does. But if I were betting on a strategy for my kids, it would be this: find something you care about deeply enough to build, and learn to use every AI tool available to make it real.

The people who figure that out — who treat AI as a force multiplier for their own passion and depth — are the ones who'll thrive. Everyone else is going to be competing with systems that work 24/7 and get better every month.

That's not a comfortable answer. But I think it's an honest one.