Alex Hormozi

I want to have agents that do work for me.

Where people fall off is that they're not training AI the way they would train a new employee.

And so they have the, I was about to say employee, they have the agent do something and the output goes back and it sucks and they're like, oh, this will never work.

Again, I will remind you, this is the worst that will ever be, number one.

Number two, if you had a brand new employee, and then you said, do this task for me, and then they gave something back to you, would you immediately fire them?

Probably not.

You'd be like, oh, I just need to train you more.

And if you think, oh, well, AI can't do what humans can do.

I really want to break that.

I want to destroy this for the people who don't get it.

And if you do get it, then maybe you need to send this to your employees, your team, because this is real.

Humans learn the reinforcement, meaning you do a thing, you get an outcome, good or bad.

If it's good, you do more of it.

If it's bad, you do less of it.

That's how humans learn, period.

And so when someone says, ah, but you know, this person has such good taste, it means that they recognized a pattern, they communicate that pattern, and they were rewarded for doing that.

And so they do more and more of it.

And so they get better and better at recognizing patterns.

Guess what's really good at recognizing patterns?

Even better than humans.