The Art of Network Engineering The ABCs of AI
19 snips
Feb 25, 2026 Mike Bushong, a pragmatic network engineer who brings operational perspective. John Capobianco, Head of AI & DevRel at Itencho and MCP proponent. They explore why automation adoption lags. They unpack LLMs, RAG, agentic workflows and MCP. They advise a safe read-only, human-in-the-loop hello world: connect AI to NetBox/Nautobot and build from there.
AI Snips
Chapters
Transcript
Episode notes
Why Network Automation Adoption Is Still Low
- Automation adoption has stalled: roughly 30% of networks have meaningful automation while ~70% remain largely manual.
- John links the gap to tooling friction, cultural resistance to programming, and enterprise inertia around automation timelines.
Augment Yourself With AI Instead Of Resisting It
- Do start augmenting your skills with AI rather than fearing it; people using AI will outcompete those who don't.
- John recommends learning to connect AI to real tooling and sources of truth as a practical evolution from automation.
Begin With ReadOnly AI Workflows
- Start with human-in-the-loop, read-only workflows like logs, config deltas, and compliance before doing live config changes.
- Use RAG/vector stores and MCP-wrapped show commands to let agents analyze logs and produce human-reviewable findings.