“People are real smart. So if you try to charge them based on tokens, they are very quick to say, oh, you've just marked up a Claude token 4x. Are you 4x as good as Claude? I don't know. Let me go try it out. You actually put yourselves in a competitive position against the models that probably no application really wants to be in. … If you're trying to mark up at all, users are very smart and you will lose on that comparison.”

Someone just left your company and you didn't get them out of all of their accounts. I just wrote a custom agent for you that will go do that. Do you want me to go turn this on? Great. Do you also want me to go remove accounts, access from these three people? Great.

So it's not just here's a red alert that someone didn't meet to monitoring stuff. It's like too many reds. I would do like me to go do this.

I'm also so curious that you said marking up a cloud token, but what about marking up like a Kimmy token or like something that's much cheaper? What's it been like deciding what to use, what's good enough versus cost?

This was something we started concentrating on at the beginning of the year is basically evals, task specific or agent specific evals. From a product perspective, we call it all the Vanta agent and it's one thing. Of course, behind the scene, it's like hundreds of agents that all have specific…

you talk to other founders, both of your era and of this new era, what are they doing here? Are they also eating the cost and just charging a software fee? What do you think the end state is going to be?