Tenex CTO Venkata “VK” Kopaka recently joined best-selling author and AI Expert, David Linthicum on Cloud Computing Insider to talk through what an AI-native security operations platform actually looks like in production, the alert overload crushing SOC teams, why guardrails matter more than raw autonomy, and where AI-driven attacks are headed. The conversation below has been edited and condensed for clarity.
David: You met the company at RSA a few months back, and Tenex has been on a tear, named the #1 fastest-growing cybersecurity company in the country. What does it feel like to have the tiger by the tail right now?
VK: It’s exhilarating. We’re in one of the most cutting-edge times of technology, and to actually lead the wave to some extent is the dream of any engineering team, the dream of any CTO. It’s a privilege to be in the middle of AI security and help lead the charge.
David: How did you get to Tenex? What were you doing before?
VK: I lead engineering and product at Tenex. Before this, I spent eight years at Google. I was one of the founding engineers on what was then called Chronicle (now Google SecOps) and worked across a bunch of Google’s security products, from Chronicle to VirusTotal. In the last eighteen months before I left, I was on the cloud and Gemini side of the house, working on some NLP problems.
Along the way, I realized there was a fantastic opportunity at the intersection of AI and security. Security is about keeping companies and individuals safe, and for decades, defenders have been on the back foot. Defenders have to get it right every single time; attackers only have to get it right once. This is the first time ever in history that the defenders have an advantage. So the idea behind Tenex was simple: put AI and security together and give the good people fighting the good fight an advantage above all else.
David: You did the same RSA walk I did. Everyone’s talking about AI. What actually makes Tenex different?
VK: A lot of what I saw was demoware, a slideshow about AI, or an interesting demo. But putting it in production, serving real analysts and real enterprises, is the biggest differentiator right now. Productionized proof is very different from a demo.
You’ll hear me talk a lot about analysts, because that’s who we’re helping. Every analyst drowns in alerts; an enterprise I work with gets about eight to ten thousand alerts a day. You’d need 100 to 200 people in the SOC to realistically triage that, and no company staffs that many. So we use AI to solve the triage, deep investigation, and response problems, and we bring it together with the right enterprise guardrails. It’s not about replacing analysts. It’s about helping them focus on the parts where their years of experience, their decision-making, and their judgment actually matter, and taking the noise out of the job.
David: What signals from enterprise customers are validating that you’re on the right track?
VK: It’s evolved over the year and a half we’ve been on this journey. Early on, it was fear: “How accurate is the AI?” Everything comes down to quality, and we benchmark in the high nineties. One CISO framed it perfectly: think about an analyst paged at 2 a.m. on a Saturday. They’re talented, but they’re not at their best, maybe forty or fifty percent. So you’re comparing forty or fifty percent of a human’s best against high-nineties quality, repeatable at any hour. That was the first validation.
The second was context. People assumed AI couldn’t hold the years of environment-specific knowledge an analyst builds over ten or fifteen years. That’s the problem we’ve spent the last 18 months on: giving the AI the right context for your particular environment. Analysts become “context engineers” and help the AI build far more context than was possible before. It’s always garbage in, garbage out, but give it the right context about the environment and the crown jewels, and the technology does an amazing job.
David: The human-in-the-loop piece struck me as unique when I spoke with your CEO. Where did that come from?
VK: Big enterprises don’t just care about autonomy and quality; they care about guardrails. Take a large financial institution: if some AI decides to isolate the most important service they have, that’s losses in the millions. So the question is, when the AI says “this is the right action,” how do we bring humans in to validate it?
Here’s the example I use. If I isolate David’s laptop, worst case you call IT, and you’re back up, not the end of the world. But if I isolate a domain controller, the implications are enormous. So we bring humans into the mix at exactly the right juncture to make the judgment calls no one else can make, backed by deep auditability of why the AI made or didn’t make each decision.
That comes from experience. Our CEO, Eric, has run SOCs. Our president was a CISO at a large company. This is practitioners building a product for other practitioners.
David: You went public with the Kansas City Chiefs win early. Why?
VK: It ties back to validation. The Chiefs are a fantastic, advanced security organization; they have a huge brand to protect. Moving from a demo to real enterprise validation matters when you’re running a security operations platform and doing triage at scale with real guardrails. We have a lot of demoware in this industry; we want to push toward real use cases and real problems.
David: Are you creating a new category, the agentic SOC?
VK: We’re part of the group leading the charge. The traditional model, hire more analysts to keep up with alert volume, doesn’t scale. We believe in bringing agentic solutions to enterprises to solve that at scale. There are other players working the space, some in the traditional way, but we’re looking at the problem in a very different way: AI-native, from both a functionality and a platform point of view.
David: What does “trusted by the Fortune 100” look like in practice?
VK: Guardrails around autonomy, full autonomy on the laptop, humans in the loop on the domain controller. Deep auditability and a compliance layer: what’s the reasoning, why this decision and not that one. And scale at the right cost. We’ve been in a “token-maxing” world where the message was just spend, spend, spend. But real enterprises have budgets and COGS justifications. At some point, is it easier and cheaper to hire a person than to spend on tokens? If the answer is yes, you’re on the wrong path. Solving the problem cost-effectively is critical.
David: How are AI-driven attacks reshaping security?
VK: One prediction: I think we’ll see as many patches in the next six months as the industry has seen in the last ten years combined, because LLMs now let people do things they never could. The rate of attacks and patches will increase by an order of magnitude, and the only way to fight that is to put a machine against a machine, augmenting your humans rather than removing them.
You’ll also see AI red teaming take off. We’re working with a big enterprise on a purple-teaming exercise: launch a red-team attack, let our platform be the blue team, prioritize and patch at speed, then rerun it to see if we’ve closed the gaps. The only way to keep up with that pace is with AI.
David: Where can people find you at Black Hat?
VK: We’re at Mandalay Bay, Banyan F on floor three, August 5th through 6th, 9 a.m. to 6 p.m. Bring us your custom use case and we’ll show you the platform live. We can get the agentic platform running in your environment in less than a week. We’re also running a capture-the-flag, so you can get hands-on and triage an alert with AI yourself. Bring me the hard questions; I’ll be around the booth.
Lightning round
Biggest misconception about Tenex?
VK: That we’re just another MDR. We’re leading a new age of what I call AI MDR, backed by a huge platform and product, but built on the belief that the right answer is humans and product together. That’s why we’re an MDR. Thinking we’re only human-driven misses the difference.
Most exciting AI capability in security over the next twelve to eighteen months?
VK: AI red teaming, I think it goes mainstream and becomes available to the masses within months. Which means you need a blue-team solution to keep up, because attackers only have to win once. I read that indie hackers are among the top adopters of LLMs; someone you’d never expect becomes an attack vector because they have a $200 subscription. You have to have that defense layer built in.
Best career advice you’ve ever received?
VK: Be close to tough problems, and close to the people solving them. That’s what I did over ten years at Google; it teaches you scale and how to solve customer problems. We’re building the same kind of team at Tenex, and we’re hiring like crazy.
Where’s Tenex headed in the next twelve to eighteen months?
VK: Running AI in your environment cost-effectively. The industry is moving away from token-maxing, so you’ll hear a lot about local and open-source models you can fine-tune to your environment. It’s a hard problem, getting the right quality at the right cost, and it’s what we’ve been working on for almost a year. In the end, it’s the outcome that matters more than anything else.
The full Interview can be watched here. Want to see Agentic SOC live? Find Tenex at Black Hat, Mandalay Bay, Banyan F (floor 3), August 5–6. Learn more at tenex.ai and check out the careers page if you like working on hard problems.