The Internet of AI Agents Is the Future. In Conversation with Vijoy Pandey

The Pulse of AI episode features Vijoy Pandey, Cisco’s SVP and GM of OutShift, who outlines a bold vision for an Internet of AI agents. Drawing on two decades of distributed systems experience and more than 80 patents, Pandey explains how agents will evolve from simple chatbots to autonomous collaborators that relieve humans of repetitive tasks and enhance creative work. He argues that this shift is reshaping the very foundations of computing, moving from deterministic code toward probabilistic, agent‑driven models. For enterprises, the emerging agentic layer promises faster decision‑making, personalized services, and new revenue streams across networking, security, observability, and collaboration.

Pandey defines an ideal agent as a proactive entity that can reason, observe, act, and continuously improve on behalf of users or organizations. To make such agents work together, he proposes a four‑phase framework: discovery, identity verification, composition, communication, and evaluation/observability. OutShift operationalizes this framework through its internal incubator model, building startups that target adjacencies to Cisco’s core businesses while leveraging the company’s sales, marketing, and support muscle. The team has open‑sourced the Agency project, providing reusable code for each phase and fostering an interoperable ecosystem that breaks today’s siloed agent deployments.

Despite rapid progress, the industry remains early in its agent journey. Most products embed isolated agents, creating ambiguity around confidence levels and outcomes—a symptom of the broader transition from deterministic to probabilistic software. Pandey stresses the need for open standards, shared identity protocols, and robust observability to ensure agents can collaborate across supply‑chain partners, developers, and end‑users. For business leaders, embracing this open Internet of Agents means preparing for new governance models, investing in linear‑algebra and quantum‑ready talent, and adopting platforms that can scale agentic capabilities safely and transparently.