Cisco CPO Defines 'AgenticOps' for Networks
Cisco's President and CPO, Jeetu Patel, has described a new operational model called 'AgenticOps.' This approach represents a strategic shift from IT 'guesswork' to continuous optimization and intent-aware security for distributed networks. The concept emphasizes using agentic systems to manage complex infrastructure proactively.
- AgenticOps is part of a broader Cisco AI strategy that includes a "Deep Network Model," a specialized large language model trained on over 40 years of Cisco's networking expertise, including materials from its CCIE certification. This model is designed to provide 20% higher precision and accuracy on networking tasks compared to more generalized models. - The system is designed to keep humans in control, with all AI-driven actions being reversible and auditable. It shifts IT teams from manual intervention to overseeing machine-driven processes, allowing them to maintain governance over outcomes. - A key component of the user experience is the "AI Canvas," a shared, interactive workspace with a generative UI. This interface unifies data from various Cisco platforms like Meraki, ThousandEyes, and Splunk to facilitate collaboration between networking, security, and operations teams. - To power the demanding workloads of AI and AgenticOps, Cisco has developed its own custom silicon, including the G200 and G300 processors, and formed strategic partnerships with companies like NVIDIA. This is part of a strategy to compete with the vertically integrated AI infrastructure offered by rivals. - The rollout of AgenticOps features is phased, with capabilities for campus and industrial networks beginning in February 2026, firewall enhancements targeted for May 2026, and data center solutions entering controlled availability in June 2026. - Cisco is extending these agentic capabilities to security operations, offering proactive firewall policy recommendations based on application patterns and automated compliance checks for standards like PCI-DSS. - The concept of agentic AI is not unique to Cisco; competitors and industry analysts like Gartner see it as the future of network operations, marking a shift from AI as an analytical tool to an autonomous actor. - The AgenticOps model integrates telemetry from a wide array of both Cisco and third-party sources to inform its autonomous actions and troubleshooting. This includes data from Cisco Networking, Security Cloud Control, Nexus One, Splunk, and ThousandEyes.
