00:00This is really interesting right now. You know, defense technology generally is still a main focus
00:05of investors like yourself everywhere. But the relationship between the technology sector
00:11and the arms of government, the Pentagon, we're rethinking it a bit. What's your latest thinking
00:16on that? Well, look, I think we've always known that these models were going to get increasingly
00:22powerful and we need to work together across the private and the public sectors to make sure that
00:29we keep people safe because whatever, you know, power we build into models and try to use responsibly,
00:35there are others in the world, unfortunately, that will use maliciously. And so it's important
00:41that folks have open dialogue. We're constantly sharing and collaborating and figuring out
00:46the best way to shore up defenses for those who need and deserve it to make sure we can always
00:52stay one step ahead of those who are meaning to do harm. It's so fascinating to have you on in
00:58particular because of your experience with, you know, when you were being purchased your own
01:02company, but Nami was being purchased by VMware. You think about what the experience you had at
01:06GitHub. But for me, it's the fact that you're on the board of Linux Foundation since all the way back
01:10to 2016. Look, our Margie Murphy, our cyber reporter, is really talking about how Mythos has been a concern
01:15for the very underpinning of the internet, Erica. How much are these models giving you pause over at
01:22that foundation or more broadly, an opportunity for some of the cyber companies that you back?
01:28Look, I'm a techno optimist, right? And I think that AI can do wonders for open source, which to your
01:35point is the underpinnings to the internet. Using these models can help us find vulnerabilities that
01:41we didn't know were there. It can help them patch vulnerabilities more quickly. So I think it's great
01:47for the world. And as you mentioned, we do have a number of cyber investments, one of which is called
01:52Jetstream that is focused on AI governance and visibility. It'll allow enterprises to understand
02:00what's happening in their organization, create a dynamic graph of AI activity across the system
02:07and put in place controls to secure it. And so I think there's a lot of opportunity both on the
02:15open source side of things to bring AI to improve open source, as well as a lot in private companies
02:22to deliver tools to enterprises to get them ready for what's to come in as AI continues to proliferate.
02:29So you're the investor and you have a conversation with the portfolio companies and they'll probably go to
02:35you, Erica, what do you think we should do? And the tension is this industry says these should be
02:40smaller open source models. That is the best way for everyone to work collectively in the
02:45cybersecurity context. Anthropic and open AI do limited access to what are much more powerful
02:52models. And they argue the limited access is a guardrail in and of itself. But it means that for
02:57researchers, bug bounty players, et cetera, it's closed to them. What is the right pathway forward?
03:06Look, I think we need to be responsible with the power of these new models coming online. And it
03:12reminds me of like responsible disclosure in cyber, right? When you find a zero day, you have to report
03:18it responsibly before you share your work with the world. I see a very similar pattern here. And I'm
03:23actually very supportive of what the model vendors are doing and trying to bring together the right folks,
03:28the Linux Foundation included in Project Glasswing, to make sure that people get an opportunity to see
03:34what these models can do, to put them to work, and to make sure that they're building up tools that
03:39can defend against them if they're used for malicious purposes. Again, I think you need a lot
03:43of different perspectives, organizations, and quite frankly, just brains around this problem to make
03:49sure that we're building the right controls to make sure that we can get all the value out of AI
03:54without sacrificing our defenses. Value out of AI. At this moment, Erica, how are other bets of yours,
04:03other portfolio companies managing to get the value? How are we seeing this being adopted within teams,
04:08embraced, and pushed forward? Yeah, look, I think the best companies are adopting AI across the stack.
04:16Ramp has had a number of fantastic posts. They're completely changing their entire culture by enabling their
04:22company to build with AI across the stack. We have companies like Railway, for example, that run all
04:29of their go-to-market with AI. So it's not just about coding, where, of course, we're seeing incredible
04:35acceleration in product shipping, evolution, new feature development, et cetera, but also the go-to-market
04:43teams are building their own tooling now with AI. I mean, I'm building my own tooling for my work
04:49and venture, right? And so I think the companies that get ahead are the ones who are going to drive
04:55AI
04:55usage throughout their organization across finance, HR, support, go-to-market, and not only code,
05:03which is obviously a huge opportunity in and of itself.
05:06you know,
Comments