 Welcome back everyone to the SuperCloud 6 live here in Palo Alto, I'm John Furrier, host of theCUBE with Dave Vellante and our entire CUBE team from the CUBE research. And of course all of our great alumni and guests and our next up is our CUBE alumni, Deepin Desai, Chief Security Officer with Zscaler, leader in SaaS, SaaS security platform, doing some real innovative things around Zero Trust and how that impacts the new role of Genevieve AI and how that's sweeping the change of applications. Deepin, great to see you. Thanks for coming on theCUBE. Hey, thank you, John, and Dave for inviting me. So Zero Trust, obviously not a well-known topic, but it changes when you start thinking about the role of data and AI because it's a double-edged sword because data is needed for good security. And so you need Zero Trust, you got to have that market, you got endpoints, you got ingestion, now you got AI engines dropping in over the top on the new architecture. So you guys are in there. So what's the status of the Zero Trust with Zscaler? What are you guys doing that's innovative that people should know about? Absolutely. So Zero Trust is about allowing businesses access to what it needs when it needs it, right? And the hard part is the only part where you should be following three key fundamental principles. So the key principles around Zero Trust security models, if I were to call out, number one is never trust, always verify. Number two is assume breach. Number three is where you verify explicitly with least privilege access. Now as part of my role, I do have the privilege of talking to global Fortune 500 SISOs and we're always discussing this topic and it's a journey. Zero Trust transformation is a journey. Here is how I would encourage everyone to look at it when they're gauging the maturity of their Zero Trust transformation. And I'll call out the innovation that we're doing across the stage using generative AI as well. Number one is your Zero Trust solution should allow you to reduce that external attack surface. Preventing the attackers to go after your assets, exploiting vulnerabilities. Think about what we're seeing right now with the VPNs where there is mass scale exploitation happening. Number two is you should be able to prevent compromise with consistent security, with full TLS inspection, no matter where your users are, no matter where your applications are, no matter where your devices are. Number three, and this is one of the most important facets of Zero Trust where your solution should allow you to prevent that lateral propagation from happening. This is where simplifying user to app segmentation with advanced technologies like deception allows you to reduce the impact of that security incident. The goal over here is to contain the blast radius to the user or the application where that incident started. Because it used to be once you got in, you're in and you could just traverse. You're saying that's part of it. But if you go back to those first three principles, GenAI really hasn't changed any of that. Now it's brought in a new dimension. Exactly. But it hasn't fundamentally changed what Zero Trust network architectures are all about in those fundamental principles. Exactly. So Zero Trust principles remains the same, but GenITV AI will actually help us speed up the adoption of the Zero Trust implementation. And I'll explain you how as well. So the final point I was gonna make is all of these threat actors are after your data. Once they're in, once they move around, they're trying to steal data. This is where GenITV AI helps with data classification and security models. When I talked about the lateral propagation piece, this is where organizations often struggle with the segmentation implementation journey. Using LLMs, using both generative and predictive models, you're now able to implement intelligent policies that simplify, that's able to sift through large volume of data based on how the access patterns look like over the past several months, establish that baseline and quickly tell the IT and security team, if you were to implement this, here is the impact, here is the security posture improvement that you're going to get out of it. So at each of the stages, if you think about reducing your attack surface, preventing compromise, using generative and predictive models, that's where you're trying to detect breach-like scenarios where a threat actor is probing, doing recon, trying to get in the environment, as well as on the segmentation. What I like about that is that the lateral movement, that's where the damage gets done. You guys have data on that. ZScale has been very successful, the stock price, your performance has been key. You've been watching that. Do well, congratulations. But now the data that you have and a platform, that's an opportunity with GenerAi. Can you explain the opportunity and what you enable? Because if you have the platform, you've got all that security in place, now the data could work for your customers. How are you see that evolving? What steps are you guys taking specifically to make that management of the data and that security paradigm work better? Great question. So data is the new currency, data is gold in this AI revolution that we're part of. One of the initiative that I'm personally driving, this is public information, Jay talked about it as well, is exactly around what you are. So we have endpoint agents. So we have endpoint visibility. We have north-south visibility using our ZScale or internet access where we're securing users going to internet-bound destination. We have east-west visibility with the whole segmentation connecting users to the private app. And then on the data prevention stage as well, we see the data that's residing in the SaaS applications, internal application. Now combining all of that telemetry and visibility, bringing it into our AI breach prediction model is something that we're building right now. So the way to think about it is... Breach prediction. Breach prediction, yes. So you said generative AI is saying this could be a breach area. This is exactly the direction we're going in. The goal over here is to combine the visibility across the chain and rich that with the intelligence. The volume of data that we see is getting close to 400 billion transactions a day. This results in 500 trillion daily signals. All of this combined with the intelligence that the team is generating is what we're leveraging to further tune those models to predict breach-like scenarios. That's pretty large scale right there. I want to ask you how Zero Trust fits in. John was asking about stock price. I'm not going to ask you about stock price. But last... I don't think I asked about stock price. I highlighted stock price. You mentioned stock price, which was kind of a head scratcher for me last quarter because you guys beat and raised and then the stock went down. But regardless, what was interesting that came out was Nikesh from Palo Alto said, use the term spending fatigue and that set off a whole wave. But Jay on the Z Scaler call and George Kurtz on CrowdStrike said, we're not having that problem. And so you don't necessarily comment on the numbers, but I'm interested when you talk to your 500 colleagues, are they seeing spending fatigue and consolidation fatigue? Why are they seeing that? And how does and will the proper implementation of a Zero Trust network architecture address that? What are you hearing from your peers? Great. So Dave, the question is two level below the pay grade of the folks that you mentioned, but I'll share my perspective. Platformization is real. And there are two aspects to it, right? Number one is where you obviously want to reduce your cost, but the most important part over here is reducing complexity. As I talk to my peers, all of them have built up a debt of so many different security tools and these are point solutions. The speed at which the attacks are happening these days, I was reviewing one of the ransomware attack that happened last month, it took 12 hours from the point of entry, exfiltrating three terabytes of data and then encrypting all the files in the environment. Now, if you have five different point technologies that are doing things in the back end and then you have correlation happening with the yet another tool and your team is trying to respond to that, the damage is already done. So, platformization is required, this is where you need to have key areas identified and that's where, for instance, Zscaler Zero Trust Exchange helps with the four areas that I mentioned, external attack surface, compromise, lateral propagation, data exfiltration. You need to have that single platform view where each of these engines are feeding intelligence into each other and you're able to take actions at the time incident is happening and not few hours later. So that, I mean, for you internally, of course, you're Zscaler, so you're standardizing in the Zscaler platform, but a platform implies that you're going to be able to evolve that over time. Do you see a risk deep in, and maybe it's okay, we kind of saw this with cloud, we had a lot of consolidation but now you've got multi-cloud and super cloud and you've got complexities there. Do you see the risk of having point products? Yes, they'll consolidate but then you'll have multiple platforms. How do you see that playing out? So there has to be an ecosystem. Like if a vendor claims to do it all, yeah, you're running into another set of problems which I won't digress into. For instance, in our world, we don't compete in the endpoint side. There are EDRs of the world, CrowdStrike for instance, a great partner. We see it there. They're doing great on their end. There are different aspects where you need to have that seamless API integration. That's equally important when you claim to be a platform, if you're not integrating well with other platforms, then then you're not essentially a platform. So again, to answer your question, you need to know the space you're playing in and you need to do it well rather than trying to claim to do it all. Getting several different point products and then again stitching it together and calling it a platform is not really a platform. But one of the things you're highlighting that's interesting is that earlier you said you're bringing in, you're enriching data with other data. The intersection of multiple data sets or data aggregation becomes a new pattern that everyone's accelerating on because now with LLMs and new foundation models, Generative AI is about bringing data to the table and then enriching it with others. That seems to be a big part of the trend. So if that continues, there's going to be an impact to customers. How do you see Generative AI impacting the customers? What's in it for them from an enablement standpoint? What is it going to enable? Yeah, Generative AI is, again, the number one piece is it's going to assist them in implementing a lot of these initiatives including zero trust adoption, where there is a fatigue, as you mentioned, because of a lot of tools. What do I prioritize? Where do I focus my energy on? With the ability of large language models to sift through large volume of data and bringing a simplified view in front of them. I gave an example of intelligent policy, policy impact analysis, generative predictive models. So it will definitely help them fast track the journey. Now, the change is bidirectional though, right? It's not just, you know, you're leveraging Generative AI to fast track a lot of these initiatives. You will need to take into account the Generative AI applications as well from the security perspective because there is risk that is associated with these LLMs. And we have seen several incidents over the past 12 months where an employee inadvertently leaked data or it could be a malicious insider. So that's where, how do you leverage zero trust to also securely enable adoption of Generative AI in these organizations? You know, Deepin, I want to get your thoughts as you talk to customers, you mentioned, you talked to a lot of other CISOs and companies. Dave and I have been covering the data area going back when we started working together 14 years ago when Hadoop was considered big data. Now Hadoop's data is key to Generative AI. There was a role that emerged called the chief data officer. And we're seeing trends now where data science and data officers are going to end up becoming abstracted away with Generative AI and the pendulum shifts to cloud ops, right? So, or DevSecOps, we're going to call it, but still data is critical. Is there a new persona emerging in the enterprise as these new teams come together to redefine or refactor zero trust for Generative AI, refactor security posture, refactor their IT? What's your vision there? What do you see happening in the field? Yeah, there will essentially be a new persona. Right now it's the CISOs and our cloud ops counterparts who are taking it on. And there is role for chief data officer even now, in my opinion, ALLMs, what goes in, what comes out, the governance around it, the security around it. A simple example that I was discussing last year with my CEO, you have these large language models. Yes, they're private. You're training that with a lot of sensitive data. What about RBAC around these ALLMs? When a CEO asks a question, and if I ask the same question or my help desk employee asks the same question to that ALLM, the type of data that's- Different answers. Exactly, right? And again, how do you train that large language model to make sure things that you don't want to be coming out is not even included as part of the set? How do you prevent hallucination? So- Well, that highlights the whole point we've been seeing here at SuperCloud 6 in this AI innovators section is that the way things were before, observability, network management, application performance management, DevOps, are all changing. It's not the one thing. They're not stovepipes. They have to work as an integrated system. That seems to be the pattern. Do you agree? 100%. Yeah, I mean, if I could apply the platformization concept on the role and personas, you're right. Because we're all, again, we're no longer the house of Mr. Doe's. We're seen as business enablers, but how do we do it more securely? And that's where those personas have to work together. So you've brought up the concerns about LLMs and hallucinations and inconsistent answers. How far away do you feel that we are from things like LLMs and other AI really dramatically changing the SECOP analyst experience? So on the security side, we still have time. It's nowhere close to replacing a security professional. Now it will help the cybersecurity professional to scale and be effective, be more efficient because it's able to scale through the data. So you're now able to make data-driven decisions in matter of split seconds because you have this thing, giving you the options and you could call out what's good, what's bad. But at some point, I do see it replacing certain tiers of what SECOP's SOC activity involves. For at least filling the gaps that can't be filled today by humans because there just aren't enough of them that are qualified. 100%. Yes. The final question for you. Give us the state of the art in terms of Z-scaler capabilities. As General AI unfolds, people start to get visibility as a certain use cases. Certain things are emerging and evolving fast on the infrastructure side. What do you guys see as clear line of sight that's stable? Where's the growth? Where's the action developing? What's your thoughts on how you see this rolling, playing out from a security standpoint? Right. So I mentioned about the use case which I'm super proud of. We're also heavily investing in co-pilots, LLMs that will enable organizations to securely adopt Genitive AI as well. Everyone realizes the benefit, the productivity, the efficiency, the efficacy. So it's not a matter of if it's, everyone is on the journey of adopting several different Genitive AI applications. So over the past 12 months, we've added and we continue to add more features, enhancements to our platform that allows organization to securely adopt Genitive AI. A simple example I can give you is you've seen cases where an employee would take financial information from internal document, try to ask a question to say, chat GPD in the prompt, you're now able to apply granular access control policy with data loss prevention engine to prevent that kind of data leakage from happening. Final, one more question just jumped in my head. I got to ask this one. What's the most important thing that you think people should pay attention to that's not obvious in the press and in the news today? What's out there that's important that no one's really talking about? What would you say? The folks have started to talk about it. I'll mention two things. One is the shadow AI aspect, right? It's one thing to define these policies and hey, you're allowed to do this, you're not allowed to do that, but having those effective controls in place that gives you that visibility and ability to enforce those security control is very important. Shadow AI is going to become more and more menacing in many of the organization leading to the breaches. The other one that... Shadow AI means what? Shadow AI is where you're leveraging unsanctioned AI applications inside the platform. Like shadow IT. Like shadow IT, exactly. In the other area? And then the other one is where we're seeing more and more threat actors also starting to use generative AI, be it wishing, you saw an example show up last month, be it fishing, be it ransomware attacks. So this is where security professionals will have to leverage AI to fight AI, right? In some ways. If you guys can get prevention, preventive, generate, I predict breaches, the bad guys are going to predict the areas to attack. Absolutely. So it works both ways. It does. It'd be great to have you on. Again, congratulations, Escalade. And thank you for being part of our super cloud journey. Jay was on our super cloud too. And great to see you. Thanks for coming on. Thank you. It's always a pleasure. Thank you. Okay. More super cloud six coverage. I'm John Furrier with Dave Vellante live in Palo Alto. Stay with us. We'll be right back with more live coverage for super cloud six AI innovators after this short break.