# HN819: Recipes for Automation – A Look Inside Eric Chou’s AI Networking Cookbook

Page: https://stenobird.com/podcast/heavy-networking/hn819-recipes-for-automation-a-look-inside-eric-chou-s-ai-networking-cookbook
Text version: https://stenobird.com/podcast/heavy-networking/hn819-recipes-for-automation-a-look-inside-eric-chou-s-ai-networking-cookbook.md
Podcast: [Heavy Networking](https://stenobird.com/podcast/heavy-networking)
Published: 2026-03-20T17:25:14+00:00
Episode link: https://packetpushers.net/podcasts/heavy-networking/hn819-recipes-for-automation-a-look-inside-eric-chous-ai-networking-cookbook/
Audio file: https://feeds.packetpushers.net/link/12486/17302798/HN819.mp3
Processing state: processed
JSON: https://stenobird.com/v1/public/podcasts/heavy-networking/episodes/hn819-recipes-for-automation-a-look-inside-eric-chou-s-ai-networking-cookbook
Duration seconds: 3709

## Resource

Network engineer Eric Chou introduces a systematic approach to integrating Large Language Models into network automation workflows. The discussion covers everything from setting up local models to using advanced prompting techniques for complex configuration tasks.

## Highlights
- Main idea: The AI Networking Cookbook provides a structured roadmap for moving from basic ChatGPT usage to building custom AI-driven automation tools
- Practical takeaway: Effective prompting requires both 'persona' (telling the AI who it is) and 'technical context' (providing specific configuration examples)
- Practical takeaway: Using 'agentic' coding assistants like Cursor can automatically ingest local project context, such as Python virtual environments and coding standards
- Failure mode: Relying on the cheapest or default models often leads to poor reasoning and hallucinations; selecting the right model for the specific task is critical
- Security takeaway: When using public LLMs, avoid uploading sensitive router configurations and utilize 'opt-out' settings in pro versions to protect data privacy

## Topics

Network Automation, Large Language Models, Python Networking, Prompt Engineering, AI Security, Agentic Workflows, Infrastructure as Code, LLM Implementation

## Chapters
- 1:00 — Introduction to the AI Networking Cookbook: Eric Chou discusses the motivation behind his new book and its systematic approach to teaching AI for network engineers.
- 10:00 — Foundations of AI for Automation: A look at the basics of LLM parameters, setting up local models, and the utility of using APIs versus GUIs.
- 24:25 — The Power of Context and Vector Databases: How providing technical context and using vector databases improves the accuracy of LLM responses in networking tasks.
- 28:50 — Crafting High-Quality Prompts: The importance of combining persona, environment details, and specific examples to get actionable automation code.
- 33:20 — Agentic Workflows and Coding Assistants: How modern tools like Cursor use local file context to understand Python environments and coding standards automatically.
- 38:05 — Model Selection and Hallucination Risks: Discussing the trade-offs between model cost, performance, and the risks of using inadequate models for complex logic.
- 56:55 — Security and Privacy in AI Networking: Best practices for using public LLMs without compromising network security or leaking sensitive configuration data.

## Actions

- request_transcript: `POST https://stenobird.com/v1/public/podcasts/heavy-networking/episodes/hn819-recipes-for-automation-a-look-inside-eric-chou-s-ai-networking-cookbook/transcription-requests` — Idempotently request low-priority transcript generation for this episode.
- read_markdown: `GET https://stenobird.com/podcast/heavy-networking/hn819-recipes-for-automation-a-look-inside-eric-chou-s-ai-networking-cookbook.md` — Read the agent-friendly Markdown representation of this episode resource.

A page view does not enqueue transcription. Agents should invoke `request_transcript` explicitly when they need this episode processed.

## Transcript

Full transcripts are not published on public pages unless there is a clear rights basis.