ATLAS · Volume I · May 2026
Memory Management for AI Agents
Four connected chapters on how agents remember, lose context, forget, and revise what they believe.
- 189 pages
- 4 chapters
- 4 code companions
Why I write
I write these books to help engineers become better at building AI systems. The most useful lessons I've learned came from doing the work: choosing an architecture, seeing where it fails, and measuring what actually improves it. ATLAS and NEXUS are my way of sharing that knowledge and experience openly, with clear explanations, honest trade-offs, and code you can inspect. I hope they give you the confidence to question defaults, test your ideas, and build systems people can rely on.
ATLAS · Volume I · May 2026
Four connected chapters on how agents remember, lose context, forget, and revise what they believe.
NEXUS · Volume I · Coming soon
The architecture of production-grade retrieval-augmented generation
An eight-chapter guide to building a measurable RAG system, from document processing and retrieval to reranking, grounded answers, and evaluation.