Just like the near-infinite replication of Agent Smith in The Matrix, autonomous AI agents are expected to pour into enterprise networks at a relentless velocity. Any doubts, just look to what Tata Sons Chairman N Chandrasekhar said at the recently concluded annual general meeting of the world’s leading IT services company TCS. In his speech he mentioned how TCS will deploy as many AI agents as human employees within the next three years. It took TCS 53 years to scale up to over half a million employees, whereas the time frame for a similar number of AI agents is likely less than 10 per cent of that period. While this will without a doubt unleash unparalleled productivity, are enterprises ready for the challenges that come along with it?According to Dr Swamy Kocherlakota, Executive Vice President, Agentic AI Security Engineering at Zscaler (a leading global enterprise cybersecurity provider), without strict guardrails, the uncontrolled sprawl of AI agents can easily become a CISO’s worst nightmare. His ‘startup’ team within Zscaler is focused on solving the ‘Agentic AI’ security problem for global enterprises.businessline caught up with him on the sidelines of the Zenith Live ’26 conference on AI Security in Vienna to understand the challenges enterprises face on the cybersecurity front in a rapidly changing world as AI permeates more and more workflows across organisations.Before we move to your solution for AI enterprises, can you first explain what role AI agents play at Zscaler?There are two sides to it: “AI for Zscaler” and “Zscaler for AI”. The first involves using AI internally to manage and troubleshoot our existing products. My mission is the second part: Zscaler for AI. We are building specialised products to protect AI implementations for organisations worldwide. We operate as an AI-native “take-off team” — a highly focused startup within Zscaler that bypasses traditional corporate friction to innovate and deploy security products rapidly. The majority of our team’s products as part of this mission were built using AI in the last six months.With employees increasingly adopting AI, what are the biggest risks companies face?The primary risk is a total lack of visibility. Many enterprises have no idea what AI activity is happening within their network. The second massive risk is data protection — specifically intellectual property. Moving data has become incredibly easy, and people are uploading information into AI agents like there is no tomorrow just to get quick summaries and solutions. CISOs must protect intellectual property from leaking through these interactions, while balancing security with the risk of becoming irrelevant if the business fails to adopt AI entirely.How exactly does Zscaler help customers secure this new environment?By implementing our products, we apply our core Zero Trust principles (treat every user and application as a potential threat until verified) directly to AI agents and workloads. Architecturally, we are uniquely positioned because Zscaler operates at the network level (verifying access between devices and the cloud), processing over 750 billion daily transactions. We decrypt every packet and observe content in real time (every piece of data passing through the network is inspected).For a customer, say a private bank, we can deploy continuous red teaming (ethical hacking simulations) to ensure their AI agents remain compliant and safe. We also implement AI guardrails. These prevent data leakage, block toxic conversations, and stop employees from feeding sensitive data to external AI platforms.What organisational changes are necessary for enterprises to successfully navigate this AI super cycle?AI is a massive super cycle, and executing it successfully requires a major cultural shift. Specifically, the partnership between the Chief AI Officer and the Chief Information Security Officer (CISO) is going to be absolutely critical. Organisations where the CISO and the AI manager collaborate closely will leapfrog their competition. Security cannot simply act as “Mr. No”. Instead, security leaders must work hand-in-hand with AI managers to enable maximum business growth while safely balancing the unique risks that AI introduces.Published on July 1, 2026