No Priors: Artificial Intelligence | Technology | Startups
00:35:02 · Sarah Guo
[00:35:02] S1: Just because it is, you know, your job to think ambitiously about what the defenders can do all day. I I wouldn't say, like, AI def... Sorry. Defenders... Security defenders are doomers in that specific term, but they're, like, generally kind of dour and, like, pretty pessimistic about the cat and mouse game getting any better in their favor. Can you can you make your best case for, like, actually, you know, if people invest correctly and we make better products and we automate many... Like, is there a version where it's actually defense dominant over the next decade?
00:35:36 · George Kurtz
[00:35:36] S0: I think it can be because if we again just with with the concepts that you know are not necessarily new or unique, but that the ability to always rapidly understand what the offense is doing and making sure that your defense is keeps up with it is going to be the key. I mean, it's sort of the modern day version of the hacking exposed book that I co authored many years ago, which was like if you show people how the offense works, the defense is going be a lot better. Now, that was a much slower sort of tempo than it is today, but I think that's going to be important, and as an industry, making sure this is a key thing for us, and I really drill this home with our team at CrowdStrike, is you've got to give as good or better capabilities to the defender than what the offense has. And if you're just slightly better, you know, just mathematically, you're still going to have an advantage. And I think it's again, it's always going be a cat and mouse game, but being really smart about it is going to be important. The only way to fight AI today is with better AI. And you know you've you've gotta use the advantages you have. The defender knows their own systems. The defender has the architectural diagrams. The defender has the code, right? So how do you put all of that into something that you can dramatically increase your probability of outsmarting the offensive AI?
00:37:01 · Sarah Guo
[00:37:01] S1: The only tactical thing I I think is an interesting idea is the idea that you could, over time, if if it is going to be mostly agent written code, you could enforce better security architecture. And, you know, perhaps fear vulnerabilities or a lot of people where, you know, formal verification has always been enormously niche. Right? But even even short of that, like, okay, why is all the software insecure? Because, you know, one one big reason is that engineers are, you know, there's a limited number of them with limited capacity, and we're all trying to move very quickly. But if you can scale at the speed of compute or your willingness to spend and agents have infinite patience, like maybe we could actually end up with more secure software over time. It's an interesting... I
00:37:47 · George Kurtz
[00:37:47] S0: think there is like the industry's at a point now where the code is probably good enough where you're going to get definitely some benefits. But again, the code was was also trained in other areas like it was trained on stuff that you know may have had issues, so you know I have my own developer license. I build my own sort of apps and things like Claude and and you know I have my own security agent that reviews the code and it's like well Claude just created the code but then Claude is now with its security app going well that code is insecure. Well you just created it like why don't you just do that in the first part? You know, maybe there's some optimization, but I think what people don't necessarily realize is that even with with zero vulnerabilities, which I hope the industry can get there, you're still going to have breaches because a lot of what we see people today doing is they're not necessarily breaking in the logging in. There's such an issue around identity, but at the end of the day, we can get better code and we can have less vulnerabilities and we can ripple that through, that'll be great. The challenge is it's such a long tail. You take Microsoft as an example. Just because of all of the the. You know, compatibility that they want to maintain going backwards, backward compatibility. You know, there's things that never kind of get fixed or modernized. And you know, that's just one important element in our ecosystem. But think about all the other software that's out there. You're not going to just fix it all overnight, so it's going be a really long tail.
00:39:15 · Sarah Guo
[00:39:15] S1: You've been running CrowdStrike for fifteen years, and before that, you were an entrepreneur. Security... I've always loved security because, like, we're never... As a business person, it's both an interesting and important mission. As a business person, like, this problem is not going away. Right? But I I feel many many CEOs of large established public companies are feeling a little bit unmoored in this period of time because so much of the value that they have built in their incumbency is is being challenged. Like Mhmm. How do you how do you, you know, think about keeping yourself up to date, keeping the organization agile, continuing to win? General advice on that?
00:39:59 · George Kurtz
[00:39:59] S0: Yeah, happy to jump into that. So first, I think you have to recognize it's a different operating model and cadence. You know, we've got thousands and thousands of people, right? And when you think about some of the AI native companies that just started, they have a handful of people. Know, Claude is running all day. They've got agents doing all kinds of stuff like that's the that's the new world, which is great, but you can't be in a not that we are, but you can't be in like the waterfall mentality of development, like things pop out a year later, you're going be out of business. So I think you first have to realize that having these massive teams and these long schedules, like that is not going to cut it. So having smaller teams that are, you know, squads or you know smaller groups that are focused on a problem, they can leverage AI, but you can't have like thousands of people show up to solve a problem. You got to have a smaller team to be able to do that, and we've done that. We've done that with like our Falcon Guardian product, and to your point, like I was personally involved, you know, at three day a week meetings. I was involved in the design of it. It's like here's what we're doing. Like change this, change that, and you got to get back to your like startup roots because it takes that level of change. Know in a bigger company and there are many companies that are much bigger than we are, and you know you just can't be slow and plotting. So that's number one is to realize that the operating model and cadence is changing and you got to change what you got to disrupt yourself internally. And then you just have to look at like, well, what are your moats? If you're a system of record, it's a pretty big moat. We're a security system of record, right? And we have everyone's data for many years. And by the way, we're net data creators. Our sensors, we call them security agents, right? But I mean, like not to be confused with AI agents, the software that endpoint agents, right? That is a net data creator. So we're creating 7,000,000,000,000 or so events like a day. And if you're constantly creating this new data stream, that is going give you an advantage because what's the lifeblood of any AI system is the data that you feed into it. I think, you know, for us, it's always like, well, what are our motes? How do we look at it?