In the AI age, the generalist wins

Niche down. Pick a lane. Specialise, or you will have a hard time getting hired. It has been the loudest career advice for as long as I can remember, and I think it is exactly backwards.

AI is the great amplifier of breadth. It does not replace the specialist or the generalist; it supercharges whoever can aim it across a whole problem. And the person it lifts highest is the one who learned a little finance, a little ops, a little design and a little code, and never married a single stack. “Jack of all trades” used to be an insult. Now it is the whole game.

Direct, don’t do

The work now is direction. Learn to point, not type.

AI doesn’t reward the master of one. It rewards the master of direction.

Picture making something happen without ever using your hands, only your voice, describing exactly what you want until it appears. The first prompt gets you close. The second and third are where it gets good. I built a small Windows app that sits on my desktop: one click and it opens the browser, logs me into the sites I need, and pulls the data back in. No vendor, no subscription, no waiting. I could build it end to end only because I can move across the browser, the login, the data and the interface without handing each piece to a different specialist.

Breadth is holding many things at once

Here is the real edge, and it is not that AI is weak at depth. AI is good at depth, and getting better. Ask it a narrow, specialist question and it will answer well, and that will only improve.

Where it genuinely struggles is the opposite. Run a business and a single decision touches operations, sales, marketing and product all at once, six or seven or eight considerations pulling against each other. AI almost never surfaces those connections on its own. If you do not bring them to it, they never enter the conversation.

That is the generalist’s advantage, and it is exactly how I see founders winning with AI right now. They are not asking it one expert question at a time. They are holding the whole business in their head, so they know which things have to be weighed together, and they feed all of them in. Breadth is not about knowing more than the specialist. It is about knowing what has to be in the room.

The middle empties out

So the one who loses is not the specialist. It is the passive middle: not building breadth, not deepening real expertise, not even picking up the tool, because the current lane feels comfortable. That comfort is the trap. When breadth is suddenly this powerful and depth is borrowable, “pretty good at one thing, coasting” is the most exposed place to stand.

The honest limit

The fair pushback is the regulated, high-stakes work: law, medicine, finance, anywhere a wrong nuance is dangerous and a general model will hallucinate a confident, wrong answer. That is real, and specialist depth there is not going anywhere.

But the move is not “so go specialise.” It is borrow the depth. Pull in a specialist AI for the regulated part, or a specialist person. And if you are that specialist, this is your opening, not your threat: get great at lending your expertise to a wide director who can carry it across the whole build. Deep expert plus wide director is one of the strongest teams there is.

Start here

Want the range on purpose? Have real conversations with AI from other people’s seats. Ask how a finance person would feel the problem you are solving. Then sales. Then operations. AI is genuinely good at showing you a view that isn’t your own, once you ask. Do it enough and you start seeing which considerations have to be weighed together, which is the whole skill.

I spent twenty years out-running specialists by moving across the whole problem instead of one slice of it. AI now hands that range to anyone who wants it, which makes it the ultimate on-ramp for the people who were told tech wasn’t for them.

So stop marrying a stack. Learn to point, not just to type.