Why Data, Not AI, Drives Competitive Advantage

Wow.

I love that story, especially given, you know, my experience with data.

Um, I love the, the, the tr- sort of track record you've created over the past eight years of-

Thanks

... of just collecting this amazing data, um, and building what sounds like it to be a very strong proprietary dataset, um, that you're gonna ...

I, I think you're just, you know, just beginning to see the opportunity of how you'll be able to leverage that dataset, and, um, that's exciting.

Um, and, and makes me wanna jump into that first topic, um-

[laughs] Thank you

... because the first topic [laughs] talks, you know, directly to that, you know, what proprietary data drives, uh, why proprietary data drives advantages in the AI era, um, which is really interesting.

It's so true.

Like, I think a lot of people overlook and, and think right now, like, most of the AI that, that came out to start and most of the AI that exists today is not proprietary data.

It's just is, you know, general datasets that are publicly available, and everybody's using that same data, um, to try to do so many different things.

But what's really interesting is when you start, um, taking advantage of proprietary datasets.

So I'd, I'd love to dig into this deeper with you.

Totally, and, um, I very much agree with you, and I don't know if you've ever done this, but, um, I really like to cook, so I've asked for the same recipe from four different LLMs.

Yeah.

Even say, "Add weird ingredients," or, "Add unusual things," and I get the same answer.

I mean, it's about 95% of the same answer.

So I think the proprietary data that you mentioned, um, is, is really interesting.

The domain expertise that we've applied on top of that data to make it, um, useful for the use cases that, that our clients want is also really important.

But, you know, every- all the LLMs regress to the mean.

So our biggest challenge was how do we get outputs that move away from the mean? How do we get unusual things? Rya stands for radical yet acceptable, R-Y-A.

Mm.

Because great ideas, great creative ideas need to be both radical enough that they stand out, but acceptable enough that an audience will take them on.

And so the proprietary data is essential to do that, because otherwise you just get generic AI slop.