I mean, some of that is going to look the same, right? So tell us as much as you can about how you can differentiate between preparation for an attack and legitimate activity, and how do you keep that false positive rate low enough that people can comfortably or confidently trust it?

Yeah, I think that is probably the hardest piece about predictive security as a whole is getting people to not only trust it, but also, you're right, the false positive piece is you're flagging something The confidence has to be high for someone to trust you, and it has to be a valid alert.

Because it's a prediction and the action is preemptive, it becomes even more important that we're not giving wrong data or wrongfully disrupting infrastructure that is not even malicious.

So I think the way to address that is spotting the most suspicious infrastructure, I would say, is not the largest challenge.

We can flag 10,000 suspicious infrastructures.

The hard part is getting it right.

so that the customer or even the public can trust you enough to utilize and flag that data.

Can you talk just a little bit about how AI is affecting cybercrime and can predictive intelligence continue to give defenders an advantage in that?