Anytime that an industry gets really, really, really, really, really good at like vertical SaaS, we know how to execute the crap out of that. We know exactly what to do. It doesn't require that much. It requires consumption and execution velocity and all this. But vertical SaaS, I think, is not the right model of how you would attack a problem that is an AI native style problem. And I think that those are where all the interesting things will be. And again, I think a lot of the tactics that we're trying out at the beginning, they won't work. It'll turn out that they only work 95% of the time, and 5% of the time they punch you in the face or something. And that means that you have to be more adaptable and you have to assume that a scrappy thing will be more important.
But vertical SaaS, I think, is not the right model of how you would attack a problem that is an AI native style problem.