
RAG, context engineering, memory, orchestration, and coding agent internals. Taught at the depth practitioners actually need.
Start learningThe premise
Written for people who already ship with LLMs and want to understand their systems well enough to fix them when they break.
Multi-agent systems
Alone, an agent is just a loop with tools.
Given a shared goal, structure emerges: an orchestrator, specialists, delegation.
Orchestration is deciding what each model sees, does, and reports back.

Curriculum
73 chapters, 640 slides, each one a screen you page through. Read a subject in an evening. Reference it for years.
Agentic memory
Every run deposits a layer. Good agents compress experience into strata and let retrieval cut through all of them.

The reader
Chapters are screen-sized slides you move through with arrow keys. One idea per screen, generous type, visible progress. All the drama stays out here; the reader stays calm.
Open the reader