Myke Näf, a computer scientist turned VC at Übermorgen Ventures, just published a free guide that explains how the AI tools we use every day actually work. Called Aiaiai.guide, it's nine short chapters that build on each other, starting with what an LLM fundamentally is (a "stateless probability machine that predicts the next word") and progressing through chatbots, tool use, autonomous agents, and multi-agent systems. Näf wrote it because he kept explaining these concepts to people and never found quite the right resource to point them to.
The guide focuses on mental models rather than implementation details. So instead of teaching you how to build an agent, it explains that chatbots "fake memory by resending the full conversation every time," or that when models use tools, "the application is what actually runs them." Understanding these mechanics helps you use the tools better and sets realistic expectations. The agentic loop chapter breaks down how models plan, act, observe, and repeat until done. It's the kind of thing that clicks instantly once someone explains it.
Näf licensed the guide under CC BY-SA 4.0 and welcomes contributions via GitHub. He acknowledges getting "considerable help from the machines described herein" in writing it. Most AI education either assumes too much technical background or stays too surface-level. Aiaiai.guide sits in a useful middle ground, accurate enough to be worth reading but accessible enough to actually finish.