online posts 26
posts

How Much Code Do Network Engineers Need in the AI Era? (AutoCon 6, Part 11)

Erika Dietrick’s AutoCon 6 lightning talk: when to write the code yourself, when to let AI teach you, and when to let it write. Why you still have to know what you’re doing, and why approval is the control that matters.

Cover: you still have to know what you're doing. A terminal shows an AI that passed 100 percent of tests it wrote itself flagged as a student grading its own homework, next to a chain from write it yourself to AI as teacher to AI writes what you understand
part 11 of 11 · series AutoCon 6
  1. 1State of automation: start with your alerts
  2. 2Digital twins at scale
  3. 3Observability and blast radius
  4. 4AI evaluation
  5. 5Automation ROI
  6. 6Leading radical change
  7. 7Build vs. buy
  8. 8Wouldn't it be cool if...
  9. 9Lightning: the PB&J method
  10. 10Lightning: 3,000 routers
  11. 11Lightning: how much code is enough
series overview →
Short answer
Network engineers need enough code to think programmatically about the network. If you can’t write working automation yet, write it yourself. If you can but lack real-world experience, use AI as a teacher. Only experienced automators should let AI write the code, and only code they already understand, because you can’t review what you couldn’t have written.

Erika Dietrick skipped the “is coding dead?” debate in her AutoCon 6 lightning talk and asked a better question: how much do network engineers need to know? Her answer is enough to think programmatically about the network. She lays it out as 11 foundational topics taught in network terms, from developer mindset and environment setup through data structures, troubleshooting, libraries, APIs and version control.

When should a network engineer let AI write the code?

Where you areWhat to do with AI
You can’t write working automation yetWrite all of it yourself. No AI.
You can write it but lack real-world experienceUse AI as a teacher, with structured prompts.
You’re an experienced automatorLet AI generate code you already understand.

Then she went through the objections one at a time. What about vibe coding? Not reviewing the code isn’t safe for infrastructure. AI agents, agent loops, a spec written by a software engineer? You still have to know what you’re doing. Skills and rules files? Useful, but in her experiment, telling a high-end model to look for specific issues gave it tunnel vision and it missed others.

The line of the week for me: an AI that passes 100% of the tests it wrote itself is a student grading its own homework.

My take: you can’t review what you couldn’t have written

I agree with all of it. Her test is the one I’d use for hiring and staffing: you can’t review what you couldn’t have written. If a team member can’t read the playbook an agent produced, they can’t approve it, and approval is the only control that matters.

The self-graded-tests point ties to the best engineering pattern I saw at the conference, from Sean Mahoney’s agentic delivery talk: tests are written and locked before the build, and grading happens outside the builder’s context. Separate the thing that produces from the thing that judges. Eric Chou’s evaluation talk makes the same argument for agents in production.

Her closing point is the one I’d tape to the wall. We’re engineers. We’re responsible for the network, not the AI. That’s the whole reason I keep execution deterministic and keep a human on the merge. Agents propose. Ansible disposes. People own the outcome.

Her iDoIt program (Level 1 is 12 weeks of video, exercises and a mentor conversation; Level 2 is self-paced with a prompt guide) is free. Worth sending to anyone on your team who’s starting out. If they’re starting from zero, the fundamentals in YAML for network engineers and keeping passwords out of playbooks are where I’d point them first.

This is part 11 of my AutoCon 6 series.