AI Training Session 1

This is the only session that goes "under the hood." Everyone has used ChatGPT or Claude in some form; very few know *why* it does what it does. That lack of a mental model is the source of nearly every bad AI interaction people have — bad prompts, over-trust, under-trust, frustration. One hour of genuine understanding saves hundreds of hours of misuse later. Crucially, this is not a how-to-use session. It's a *how-it-works* session. Every subsequent session cashes in on the concepts laid down here. To keep it from feeling dry, the session **opens with a live demo of the worked example** (the Weekly Digest Agent) and **threads it through the key concepts** — context window, hallucination, and the LLM-as-OS teaser all reference it. People leave with both the mental model and a hunger to build the thing they saw.