Start here
Start here.
New to the notebook? Read these eight pages in order. They run from how a machine learns to how an agent acts safely. Each one is told three ways.
The path
Pick a voice in the header before you start. Kitchen table, Whiteboard or Boardroom. You can switch at any time.
-
Machine Learning How a machine learns a pattern from examples. Every later page builds on this one.
Concept · shelf one -
Large Language Models What sits inside a chat model. Why it can sound right and still be wrong.
Concept · shelf three -
Retrieval (RAG) How a model looks up facts before it answers. An answer is only as good as this step.
Concept · shelf three -
AI Agents A model plus a loop plus tools plus state. The model is the smallest part.
Concept · shelf four -
Evals & Hallucination How to tell if an AI system works. Where the tests fool you.
Concept · shelf five -
Guardrails & Responsible AI How to keep an agent inside the lines. What a real control looks like.
Concept · shelf five -
The daily page One idea each morning, told three ways. Come back for the next one.
Daily -
The Atlas of World Models Where my research goes next. Machines that build a small copy of the world and plan with it. The Atlas has its own Start here path.
On the Atlas
Done with the path? The Concepts page lists every idea by shelf. The topic pages tie them to the daily entries. Concepts and topics