Best case scenario, like the bad money is just quiet. Like the worst case. They're actually a distraction and pull you off the rails.

One of the biggest pitfalls, I think, as an entrepreneur is you want to showboat with your investors and you want to spend. And it's like you kind of handle your investor as a constituent to everything is good. And it's so toxic in the end because you're basically creating a culture of misinformation in a way to yourself, to your investors. And it's a terrible reflection of the relationship with the investor. They must not be very valuable to you, or you might not be able to handle tapping their value if you're doing this. And frankly, so often, and I've been guilty of this in the past, you know, like I'll manage a board meeting, I'll manage a deck, but I'm really dealing with the hard things because to really do the hard things is to be vulnerable with those people, to put yourself on the line. And it's hard. You'd rather just maybe spin it and be like, hand wave it. And we're all very good at that. We're all pretty good. If we've managed to raise millions of dollars, we're very good storytellers, one way or the other. And that means you can tell a nice story, whether it's to your board or to whatever other meeting you're doing. And I think it's really a blind spot. And one of the things I learned over the years is stop doing that because it doesn't help in the end.

I've kept you way longer than I promised. Let me end with three. Two of them are going to be a little poking at things, two questions poking at things. And then one specifically for David. But so the first poke is we've sort of danced around this idea that, well, why can't a general purpose model do this? But I mean, writ large, that's what everybody's thinking. Well, why can't Claude come out with this product tomorrow that will eat your lunch, right? So you mentioned Terroir, which I loved, which I think Nick knows, Chris and I has that's been our investment thesis as well. Can you make an argument or underline your argument again for why this sort of distillation to use that word of this expertise in a narrow field or a targeted sort of thing will beat the, oh, well, we have all of human knowledge ingested in our model. I'm

going to take the qualitative answer and then I think you can do the quantitative. So here's the qualitative, Ryan. The moat here is trust. Like you can't point to something like Anthropic and say, oh, there's no way they could like technically, technically approach what we're doing. Of course they could. They have a humongous amount of capital and thousands of talented people. What they don't have is trust. You've got a world of top-tier knowledge creators, right? The doctors and so on of the world. And if you ask them, how do you feel about Anthropic and OpenAI? They're angry as heck. Why? They stole my work. They scraped it. Then they pretend to be me. And then they give people misinformation that I don't believe in. I'm so mad. I hate them. So it's hard for those people to go to that community and say, you know what, trust us. We're going to build these small models trained on your knowledge. We're going to create this marketplace.

And I could try, but