The Everything Feed - All Packet Pushers Pods HN790: From Rule-Based to Goal-Based: Rethinking Autonomous AI Operations (Sponsored)
18 snips
Aug 1, 2025 Omar Sultan, Cisco's Director for Product Management of Automation and AI, and Javier Antich, the Chief Mad Scientist for AI at Cisco, dive into the revolutionary shifts in network operations driven by AI. They explore the transition from rule-based systems to goal-based autonomy, enhancing decision-making and adaptability. The discussion covers the necessity of knowledge graphs for AI coherence, security risks tied to AI autonomy, and practical strategies for adopting agentic AI within organizations. Their insights pave the way for a future where AI and human engineers can thrive together.
AI Snips
Chapters
Books
Transcript
Episode notes
Broader AI in Networking
- AI in networking includes heuristics, algorithms, machine learning, neural networks, and LLMs, not just LLMs.
- Some AI like algorithms and ML are more dependable and easier to develop than LLMs for certain tasks.
AI is Probabilistic and Error-Prone
- All AI, including LLMs, is probabilistic and prone to errors.
- Trust building and managing acceptable error levels is crucial when deploying AI in network operations.
Human and AI Mistakes Parallel
- Humans by nature make errors in network operations just as AI does.
- Risk management applies to both human and AI decisions in network operations.
