Synopsis: Aviatrix CTO Anirban Sengupta dives into the networking and cybersecurity implications of artificial intelligence (AI) agents that could one day soon number in the millions.
In this Techstrong AI interview, Anirban Sengupta, CTO of Aviatrix, discusses the emerging challenges in networking and security that come with the rise of AI agents. These agents—autonomous programs that perceive, analyze, act, and learn—are expected to number in the millions and operate across cloud, on-prem, and edge environments. Sengupta outlines how this proliferation will dramatically expand the network attack surface and strain infrastructure, requiring high-bandwidth, end-to-end encrypted connections and deep observability to manage the interactions and traceability of agent activities.
Sengupta emphasizes that many organizations are overly focused on protecting data, while overlooking the risks to models and agents themselves, which represent valuable intellectual property. He warns that the attack landscape is already evolving, with sophisticated threats exploiting vulnerabilities daily. To mitigate this, Aviatrix offers a unified platform that embeds security directly into the network fabric—eliminating the need for bolt-on security and enabling secure, scalable multi-cloud and edge connectivity without overhauling existing infrastructure.
Finally, the conversation turns to strategy. Sengupta advises enterprises to build infrastructure with embedded, layered security from the outset and adopt repeatable processes that allow developers to innovate without friction. He stresses that edge deployment is becoming critical, as AI agents often need to operate close to where data is generated. With models trained in the cloud but executed at the edge, a secure, distributed network architecture is vital for agility, performance, and protection in the AI-driven future.
