Every operating system now ships its own small AI model, and none of them talks to the others. This introductory course explains — in plain terms, but in real depth — what Lybo OS is, the problem it solves, how it extends the model-agnostic LyboAI Edge Runtime, how one governed agent is routed across Gemini Nano, Apple Foundation Models and Phi Silica (with a fully-offline sovereign fallback), how structural guardrails keep it honest, and how the A2A mesh connects a fleet. No prior on-device AI experience is assumed; by the end you can explain Lybo OS to an engineer, read a routing trace, and run the live demo.