Agentic AI for Networkers - BRKOPS-1327

As Yuval Noah Harari notes, humanity's unique ability to collaborate in large groups is what sets us apart from all other species. Today, generative AI is enabling a similar leap - enabling software to collaborate with software in natural language - similar to humans: Agentic AI. Agentic AI refers to systems composed of multiple AI agents that can autonomously reason, plan, collaborate, and act in pursuit of complex goals, with minimal or no human supervision. This session explores the past, present, and future of agentic AI, with a strong emphasis on practical, demo-driven insights into how these systems support and enhance networking use cases. We begin with a review of current AI assistants in networking, then transition to a deeper dive into how agents can coordinate, reflect, leverage tools, and work together to solve problems. Demos will showcase agentic AI automating workflows, customizing user interactions (e.g., AgenticOps with AI Canvas), and even reasoning across unfamiliar data schemas in cross-domain and cross-organizational troubleshooting scenarios. The session concludes by exploring distributed agentic AI - what some call the Internet of Agents. We'll discuss both the infrastructure and semantic layers: how agents are discovered, connected, secured, and observed using protocols like A2A, MCP, and frameworks like AGNTCY.ORG; and how they simulate human-like cognition through structured reasoning, group memory, goal alignment, and interaction patterns. Attendees will leave with a clear understanding of the current landscape of agentic AI, its real-world applications in networking, and where this transformative technology is headed next. Speaker: Frank Brockners, Distinguished Engineer, Cisco - Distinguished Speaker, Hall of Fame Elite Speaker Event: 2026 Las Vegas Technical Level: Introductory Technology: AI Session Type: Breakout Track: AI, Networking