Henar Lab #001 — NetOps AI Assistant
A local, free, and privacy-friendly AI assistant for network operations (NetOps). It answers questions about your network by combining:
- RAG (Retrieval-Augmented Generation): answers based on your own network documentation, not guesses
- Tool Calling: checks live device status via real ping requests
- Hallucination control: honestly says “I don’t have that information” instead of making things up
Built entirely on top of Ollama — no API key, no cloud costs, runs fully on your own machine.
Why this exists
Most AI chat tools either don’t know anything about your specific network, or confidently make up wrong answers (IP addresses, statuses, etc.) when they don’t know. This project is a small, transparent example of how to combine local LLMs with real data and real checks to avoid that problem — built as a learning project by a network engineer moving into AI/NetOps tooling.
How it works
The assistant reads a plain-text file describing your network devices, checks live device status with real ping requests, and combines both with your question before asking a local Ollama model to answer — so it only responds from what it actually knows, never from guesswork.
Limitations
- This is a learning/demo project, not a production monitoring tool
- Device status checking currently uses ICMP ping only — it won’t detect devices that block ping but are otherwise reachable
- Answer quality depends on the local model you choose (default: llama3.2)
About
Built as part of a 30-day self-taught Python-to-AI learning journey by a network engineer transitioning into AI/NetOps tooling, under Henar Lab.
