Together with DJ Sampath, SVP of AI Software and Platforms at Cisco, we take a deep dive in the evolution from AIOps to Agentic Ops. The main focus of the discussion is on AI Canvas, Cisco's latest innovation for network troubleshooting. Unlike traditional chatbot-based AI tools, AI Canvas leverages a purpose-built deep network model trained on 30 years of human network engineering interactions, the company claims.

The conversation explores three core principles of agentic operations: bringing data across silos together, enabling multiplayer collaboration, and using purpose-built models for specific domains. Sampath details how AI Canvas aims to transform IT troubleshooting from a reactive, dashboard-heavy process into an intelligent, agent-driven experience that autonomously debugs and resolves network issues.

Key takeaways:
• Agentic Ops represents autonomous AI agents executing tasks end-to-end, not just chatbot interactions
• AI Canvas uses a deep network model trained on real human network engineering data, not just synthetic data
• Three pillars: unified data silos, multiplayer workflows, and purpose-built models
• AI Canvas creates dynamic, generative UI widgets that it populates based on troubleshooting context
• Starting with Meraki and Thousand Eyes integration, expanding to Cisco Cloud Control
• Future expansion to include third-party vendors through MCP servers

Chapters:
0:13 - Introduction
0:25 - Understanding AI Canvas
1:43 - From AIOps to Agentic Ops
6:43 - Three core principles of Agentic Ops
8:11 - Deep Network Model explained
9:24 - AI Canvas in action
13:07 - Automation and workflows
16:23 - Prerequisites and getting started