IT Visionaries

The episode packs in a coherent, usable framework - read-only documentation → testing → compliance → triage → fully agentic - with several non-obvious sub-ideas (Git diff for network state, mean-time-to-innocence framing, routing table mind maps). The nostalgia tangent about dot-com networking and the 'plumber' banter eat several minutes without adding substance.

“now you've got mean time to innocence, mean time to detection, mean time to remediation, all from a project that was just having an agent do documentation for you”

“I asked a forum at the network on AutoCon in Texas, a room full of about 500 people, all of them some of the best and brightest in the industry. Who here knows if every interface on the network is clean, right? Meaning error-free. Not one single person put their hand up”

The IDS/IPS analogy for agentic AI risk-gating is the freshest framing in the episode and genuinely illuminating. The read-only-first phased approach is practical but widely discussed in network-automation circles; most other ideas (Git for config state, human-in-the-loop) are established patterns rather than contrarian takes.

“It's a little bit like, you know, the idea of intrusion detection versus intrusion prevention systems”

“if the remediation is to shut an interface and that interface is on this classification of device in this location, you can go ahead and shut the interface”

The guest is a genuine hands-on practitioner who has built specific tooling (MCP server on markmap.js.org, PyETS integrations) and speaks with customers on real deployments; they are not a thought-leader tourist. Their identity is never named in the transcript, and the 11-minute format limits how much depth can surface.

“I wrote a mind map MCP server built on markmap.js.org”

“I've been giving this a lot of thought because we have customers that want to do it and I'm trying to be a partner and a guiding light”

The guest names concrete tools (Netbox, Notobot, Infograph, ServiceNow, Postgres, Git, PyETS, markmap) and specific CLI commands, which is well above average for the genre. What is absent is any hard performance data - no dollar figures, no measured time savings, no customer case studies with outcomes.

“Sources of truth populating say Netbox or Notobot or Infograph with your inventory using AI to do that. That used to take weeks of human effort”

“Store all this in a Postgres database or store it in a Git repo”

The host is nearly invisible in the transcript - contributions are limited to brief affirmations and zero probing follow-ups. There is no challenge to any claim, no request for evidence, and no attempt to stress-test the guest's framework, making this essentially a monologue with a nodding interlocutor.

“That's a really good analogy. Yeah, I like that.”

“I know. I know. It's the more you learn about it, the scarier it becomes, right?”