Grant Harvey

AI agents have been the hot topic of 2025.

We only really started to work out how the hell to use them in January.

Like this is new and it's really like experimental.

So people are like discovering like this is why this is why the whole deep seek R1 thing came out.

They were very excited because they discovered a new way to use reward modeling data much more efficiently.

And that had a massive potential impact in the way you can train.

I mean, it does.

Distillation means that like it's a lot easier to pull stuff that already exists and make it easier rather than to do new stuff.

But like it's all it's new.

It's discovery.

I think it's why it's so exciting.

It's like there hasn't been a scientific field that has felt so relevant and current for a long time.

Corey, Corey, it's so tiring.

Let me tell you.

It's so... Because you're just like, oh, should I read this paper?

Is this important?

Do I need to kind of... And I will say, like, this is where...

People talk about an AI bubble.

My personal perspective, you could pause model development today and the consumer wouldn't notice for five years because it's about the app player.

It's the piece that sits on top.