SPEAKER_04

And then we tell ChatGPT, now create 1,000 questions like this.

So, you know, it's artificial data or artificial questions.

We curate those to make sure that they're good.

Then we do 100 hypotheses, and we create thousands of types of hypotheses, et cetera, in the same four categories.

tests that you might run.

So now from A to Z, we have an agentic AI that you give it raw data, it knows what to do with the data, it then generates hypotheses for you, and then it literally tells you the kinds of experiments you should do next to prove or disprove the hypothesis from the raw data.

Yeah.

I mean, we're not working with them directly on it.

And we first thought to turn it into a company, because that's kind of one of the things we do in my lab, because I've always thought that it's important to give back to the taxpayer the money that they've invested in us.

And the best way to do that is commercialization.

I'm totally unapologetic about that, even though that got me in a lot of trouble at Stanford in the early days when making money was, commercialization was evil.

and even at Stanford.

And so I think that that's an important process because scientists are good at asking maybe the questions and coming up with solutions, but scientists aren't the best at commercializing it and turning it into a product that can be used or testing it in large communities.

So the AI that we developed, we thought, okay, well, maybe we can do this.

We thought, you know what?

AI is moving so fast.

Why don't we just give this to the community?

Why don't we open source this?

We can use it for maybe specific targeted purposes, but we're basically going to publish the whole thing on GitHub to let other people use it.

Because we've seen other people make claims about stuff that they've already made, and it's like, oh, ours is better.