A friend and I got into it a couple of days after ChatGPT came out. Everyone around us suddenly wanted to learn how to train models, or build apps, or launch a startup on top of this thing. I said all of that is real, you can build a career training models, you can build apps, you can start a company. But look at where the money is going: the big labs are putting billions into frontier models, and there's no world where a small team competes with that directly. Training models at the frontier level just isn't a lane most people can enter.

What follows from that is the opposite problem. Once everyone has access to the same base models, the barrier to building an app or a SaaS on top of them drops close to zero. So that layer gets crowded fast too, a lot of people building roughly the same wrapper around the same API.

The lane that stays open is teaching. Not building the models, not even necessarily building the apps, but being the bridge between what's getting built in AI and the people who need to learn how to actually use it. Low barrier to entry, and the tools themselves become somewhat disposable, if one goes obsolete you teach the next one, the skill of explaining and onboarding people doesn't expire the way any single tool does.

My friend pushed back: why would anyone follow me over someone already popular talking about the same stuff? My answer was that people don't follow you only for the information, they follow you because they see themselves in you. In my own experience, the people who related most to what I shared were people who looked like me or came from a background close to mine. That's not a content strategy, it's just how trust actually forms.

There's a second-order benefit too. Being visibly good at teaching AI is also a credibility signal in front of recruiters, business partners, and potential clients, it's proof you actually understand the tools rather than just having used them once. If you run an agency, that credibility is what brings the clients in.

I still think this holds. The barrier to entry is low and the ceiling is high, which is a rare combination.