The startup unit inside Zscaler working to secure AI
Editor's Note: This is Part 1 of How AI is Changing Enterprise Security, a three-part video interview with Swamy Kocherlakota, executive vice president, agentic AI startup at Zscaler, produced jointly by Zscaler and FastForward. This installment looks at Building for an AI world.
Swamy Kocherlakota joined Zscaler in January to undertake a challenging job: build a startup unit dedicated to developing AI security products inside one of the largest cybersecurity companies in the world.
He said that Zscaler founder and CEO Jay Chaudhry (whom we profiled in this space last month) wanted someone to come in and operate with the agility of a smaller company. "One of my founders' beliefs is that big companies will slow things down, and in order to take advantage of the speed, you have to launch these takeoff teams, empower them and then allow them to move as fast as possible," Kocherlakota said in the video interview.
That ability to move faster is essential in a rapidly shifting technology landscape, and the real mission of his group is building tools that protect AI. "AI is moving very fast, and customers' needs are moving fast. Every day there is a press release or an update. The question really is, how do you build products that allow our customers to protect AI and take advantage of AI," he said.
Part of the reason his unit exists is that they are seeing a lot of projects stuck in pilots, and he thinks that's at least partly because companies don't trust the AI enough to move it into full production. He likens unprotected AI to a race car with key safety parts missing.
"You know, the reason why a lot of things did not go to production and still stay in pilot is because [it's like] a race car without brakes, security – and what we are doing by protecting AI, is giving the car the brakes so it can actually drive safely," he said.
By looking at advanced models like Mythos before they get to customers and bad actors, his team can start building products quickly and prepare for these new models. "Now that that model is available and it's available to defenders first, we were able to understand how to protect our customers," he said. "And now over time, everybody else will have access to it. Now organizations have the opportunity to prepare themselves before the model is widely available, especially for the attackers."
Disclosure: Zscaler paid me a fee to produce these videos in partnership with FastForward. As with every interview I publish or moderate, the conversation and editing were subject to the same strict editorial standards that govern all of my work. Zscaler did not get to see this article prior to publication. See my editorial guidelines on the FastForward About page for more information.
