The Agentic Enterprise: a book about AI agents written for managers It is half written and still the clearest on the question that matters: which processes are worth handing to an agent and which are not, and how you contain the damage when it gets one wrong. AI 22 sept 2026
RAG with Python Cookbook: seventy-one recipes, and the ones I actually used A cookbook. You do not read it front to back, you open it at the recipe you need. Here is which ones I opened and what decision each one changed. AI 22 sept 2026
Hands-On RAG for Production: what breaks when the pilot goes live One of the first books written after answering questions over your own documents stopped being novel and started failing inside real companies. It changed what I promise and what I measure. AI 22 sept 2026
Wave-based semantic memory: the flaw it names is real, the fix is not there yet A preprint by an independent researcher proposing to replace the measure behind today's meaning-based search. I kept the problem it exposes, not the solution, and the difference is the point. AI 8 sept 2026
Vector Databases: the answer is usually the database you already run A book defending an unfashionable position: you do not need a new database to search by meaning, an extension on the one already in production will do. I agree, and it still has two problems. AI Architecture 5 sept 2026
The ROI of AI in Financial Services: how to read a report that is also a sales document A survey of five hundred-odd financial executives, published by the company selling the technology being measured. The numbers need the fine print beside them, and the fine print is there, which is something. AI Strategy 20 ago 2026
The Generative AI Career Masterplan: two books glued together, and only one was needed Half of it is a technical primer other books do better; the other half is a map of roles you will not find anywhere else. It is worth it for the second, written by people who hire for those jobs. AI Strategy 14 ago 2026
The AI Act Handbook: the European regulation does not apply to me, and I read it anyway Nearly five hundred pages by lawyers and consultants on how the European AI regulation is complied with, sector by sector. Uneven, machine-translated, and still the best checklist of questions I know. AI Strategy 5 ago 2026
Structure of a prompt: four slides that fix the order almost everyone gets wrong Not a book and it does not pretend to be: four images with ten numbered elements. It still fixes the commonest mistake in asking a model for something, which is not what you ask but in what order. AI 30 jul 2026
My own reading path for an AI architect, and what building it taught me Three pages I wrote to settle what to read and in what order. It is here because building the list taught me more than several of the books on it, and because it has a flaw worth admitting. AI Strategy 21 jul 2026
Roadmap to Become an Agentic AI Engineer: flashcards, not a roadmap Twenty-three pages, ten sections and a hundred questions with model answers. The only part that outlives the year is the learning order; the tools it names are already moving. AI Strategy 14 jul 2026
RAG Evaluation & Testing in Production: an interview guide with a production cover Ninety-three pages built on one template repeated twelve times, half of them interview questions. Two chapters survive, and those two justify the afternoon it costs. AI 26 jun 2026