An interactive, in-browser course in the disciplines of working reliably with AI on legal material — for legal researchers, not developers. Nothing to install: the code runs in this page, and you learn by changing inputs and watching what moves. Each lesson distils one reusable discipline — counting tokens, grounding a claim, convening a diverse panel — so you leave with a working cognitive toolkit, not a folder of notes.
- L0 Foundations
- L1 Getting documents in
- L2 Finding what matters
- L3 Summarising reliably
- L4 Working at scale
- L5 Reasoning across
- L6 Improving the loop
Resources & further reading — the evidence behind each lesson, with an
honest mark of how far it reaches.
Going further — take-home experiments that reproduce each lesson's
finding in a live Claude or Gemini, once you have banked the skill.