Transcript
And it's been a while since we talked, but AI is taking off. There's all kinds of challenges around data, managing it, securing it, especially after Anthropx Mythos model. What are you seeing? I think AI is coming into the market and really starting to accelerate. And the metaphor I have in my head is that, you know, first you invent the car, then you make the car go fast. Then you start crashing the car a few times when it's at high speed, you realize it's quite dangerous. People can get hurt, injured. And then you invent the seatbelt and the airbag and regulate them and make sure that people are doing it securely. So I think when it comes to enterprise AI adoption, we're at that spinning up stage where the car, AI, is starting to become faster and faster in the enterprise, accumulate more speed and also become a bit more risky and dangerous in that sense. So what are the seatbelts? What are the airbags here? Seatbelts and the airbags are complex, especially since the AI itself is changing almost on a weekly basis here. We're seeing new capabilities, new models, new operating models for how AI is being leveraged in the enterprise. Sayera, from our perspective, leveraging our strength in data and identity, is really looking at how the enterprise uses AI in two different lenses. The first one is the workforce aspect of it. As an employee in the enterprise, what AI tools am I using? How am I working with those AI tools? And how are they changing the way I work and the way that, you know, the organization needs to provide security for me? Maybe data that I used to be able to access but never actually did. Now that I have a copilot, I get access to that data almost immediately. And suddenly, HR information, financial information, my boss's salary, and, you know, people's HR reviews become much more sensitive information that needs to be protected. In the past, it was protected a bit by obscurity. Today, you actually have to make sure that these things are configured correctly. Otherwise, people are finding all of these, you know, data points across the organization so much easier when it's an agent looking for them and they don't have to do it manually. So does red teaming just become a much more regular part of what happens in every organization? Just security through trying to breach oneself? I think that the overlaying narrative of the overlaying change in security is that in the past, humans protected from humans. And in the future, agents will have to protect from agents. There's no way at which a human defense system can keep up with an agentic attacker. And that attacker can be external, you know, can be somebody using AI, using agents to infiltrate the enterprise and cause it harm or, you know, steal information. But it can also be insiders that are using AI for business purposes. And that AI just operates very differently than what a human would do naturally. It's much faster. It can consume huge quantities of data. It can try to access every and any system that it can. So the game is changing. We're moving to a world where you have to streamline the defense to become agentic in order to continue to stay ahead of the agents that are being used to get access to the wrong information, to expose wrong information. And that is the macro change we're seeing in cybersecurity. That cybersecurity has to become an agentic AI driven effort. And I think as part of that, that touches almost every part of the stack. And when we look at Mythos, Mythos is with all of the excitement and all of the interest and all of the fear. It's just the first chapter in the book, right? Like this is the first chapter in the book of how AI is going to disrupt cybersecurity as it's been done historically, and how organizations are going to need to change and uplevel and streamline their processes and make them automated and agentic in order to be able to keep up with this much faster threat that is emerging. I continue to find a lot encouraging in how Mythos has rolled out. We've seen stories over the past, especially decade of these enormous breaches into various important institutions and how many customer records were lost. Imagine if Mythos came out, and it weren't Anthropic, giving access to these essential companies and institutions upfront. If it had been some adversary or some hacking collective who had come up with it first, I mean, everybody's house would have been looted, right? I mean, this would have been mass theft on a global scale. So I think we're very fortunate that all in all, it seems like AI is still in the hands of the good guys, or at least the cutting edge AI is in the hands of the good guys. I don't know if we can count on that being the case forever. And I don't know if what is today, let's call it differentiated AI is going to remain differentiated and not become commoditized. So if Mythos today is ahead of the pack, in terms of the cybersecurity vulnerability research capabilities, is it going to be ahead of the pack in a year? Right? Or is that going to become fully commoditized available through any model, any system, and thus available, not only to the good guys? It becomes a lot more dangerous, I imagine, when it runs locally. Right now, if the most advanced models require massive, expensive infrastructure to run, well, there's only so many places that you can run that out of and, and for so long. But if you can do this on some kind of a cluster that's harder to identify, if North Korea gets the capability, etc. Well, then that's different, right? Yeah, yeah, I agree. And I think, you know, we saw, we're seeing a lot of interest around the world about that question of who gets to run the cutting edge technologies, who gets access. And I think that the ability to buy time for the enterprise and government to really set up their defenses in the right way, and build these agentic security systems, upgrade their existing stack, I think the ability to buy those organizations time is crucial. Okay, so how is this reality affecting Saira's strategy and product right now? So for us, primarily focused on the enterprise space and supporting enterprises through this transformation, I think that Saira is undergoing an immense change from being the data security company to becoming the secure AI transformation company. And that is exactly what our customers are coming to us with. You know, they're getting immense, immense pressure to say yes to AI inside the organizations, for very good reasons, right, for all of the opportunities we all understand. But they know very well that their security systems are not there yet. And they need to get them there in a very short timeframe. Right. And that is where we're seeing this immense pressure, immense urgency coming from the market from the practitioners that are telling us we need your help. And we need it today. Find a way to get us there, find a way to uplevel our security and uplevel our systems and processes, and make sure that we are ready for this transformation that is happening. And we can say yes, without completely, you know, letting the horses out of the barn. When it comes to our space and data, the whole interaction with data is fundamentally changing. Every system we've had in the enterprise was built primarily for humans and applications interacting with data. And agents interact with data in a very, very different way. They do not rest, they have no limit, they're not hardcoded to a certain pattern. So their ability to search through the enterprise data landscape, and get to every piece of data that they can access, whether they should or shouldn't, is completely different. And this is creating tons of exposure for organizations, putting them in a situation where they're very, very, very concerned. And we've seen people roll out copilot, and then roll it back. Because people are finding things that you just did not feel comfortable with, right, like, and they had to roll back these initiatives, get their house in order before, and then approach them again. The other huge concern around agents is production outages, the ability to impact existing systems, we saw the PocketOS incident a couple of weeks back. So this is another topic that the world just doesn't know how to comprehend yet. And like, we've gotten used to putting our credit card in the ATM and getting cash out, you know, to be able to be able to use these systems with almost, you know, 100% reliability all the time. Well, these agents do make mistakes. And if they have access, and if they're involved in the production lifecycle, it requires a really high level of maturity to be able to mitigate the risk of impact, and reduce that, the likelihood that these production outages will occur. And so how is this demand affecting your annualized recurring revenue? So we're seeing immense growth in our revenue and in our pipeline. You know, when you look at the enterprise sales, you have your leading indicators, and then you have your lagging indicators. And so, you know, in every leading indicator, we've managed to outdo our expectations this quarter, very, very significantly, like 150% to the goal and beyond. And that is showing us that the market is coming into this project, into this type of work. We're seeing the priority from the market around AI security and data security, and the urgency to take action at the top of mind for every CISO and every organization today. And I think that there's also a strategic imperative here, right? Like when you look at every organization right now, obviously, they care about security, and they don't want to take more risks than they have to. But they also want to reinvent the enterprise as an AI native enterprise. They want to find the right partners and the right foundations to set them up for success, not just in the current project, okay, let's adopt this use case or this use case. But how do we make sure that we are much more adaptable, much more at ready for the next use cases that are coming our way over the next two years, the next five years. And that is where Sierra really shines. Because not only are we allowing our customers to deal with the AI use case on the table today, right, be it we want to run cloud co-op on the endpoint, or we want to have a chat GPT enterprise access all of the enterprise data, whatever the use case might be, but setting up a much stronger foundation on data identity and agents in order to be able to be much more reactive and much more adaptive to the new use cases that are popping up tomorrow. Makes sense. Jotun, thank you. Thank you, John.
