Heavy Networking · Packet Pushers

Now I Understand. You Mean an AI-Safe Zero-Trust Network Automation Approach. (Sponsored)

·1 hr 12 min·5 clips
The network is the nervous system of the business. You do not get five tries at this. It is like going in for a knee replacement and coming out missing an arm.
David Gee, CEO of Curvium, joins Ethan Banks and Drew Conroy-Murray for a deep technical conversation about AI's role in network automation. Starting from his background in embedded electronics and process control, Gee draws on decades of automation experience to argue that LLMs are fundamentally non-deterministic and therefore dangerous when applied directly to network configuration. He shares specific failures including an extended battle with an AI coding assistant that could not move a CSS element correctly across multiple context windows. The core framework proposed: LLMs should operate at the top of the stack for intent parsing, process discovery, and knowledge retrieval, while battle-tested deterministic automation handles actual device changes. Gee introduces his zero-trust approach using NATS messaging to create scoped, directional procedure calls between management systems and network devices, eliminating the need for broad API access. The discussion also covers the evolution of vendor management platforms, the three tiers of network operations (DIY, premium vendor, managed service), and why network engineers must evolve from CLI operators to system architects who understand both business logic and packet forwarding.

As heard by us

AI belongs in controlled workflows, not as a shortcut past production discipline.

Heavy Networking sits in the practical middle between network automation and the newer push to hand more of it to AI. Ethan Banks and Drew Conroy-Murray keep David Gee anchored in operator reality: source of truth, probes, permission gates, temporary access, and the habit of…

Read the full review in PlayNext →

Why you'd press play

You want AI in network automation without letting it loose on production.

Read the full recommendation in PlayNext →
Listen to the show on