JC Marin
Building AI That Knows Your Data
Your organization already has the answers. They’re just hidden away in PDFs, spreadsheets, old contracts, and forgotten folders. There are also documents that no one knows how to search for. In Building AI That Knows Your Data, you’ll follow the team at the fictional Northwind Industries as they build a working system. They start from scratch to create a Retrieval-Augmented Generation system. Along the way, you’ll learn how to extract and clean real-world documents. You’ll also see how to break documents into useful passages, create semantic embeddings, measure similarity, perform vector searches, and store everything properly. All of this is kept in a reliable index. This book isn’t just a bunch of code snippets or abstract AI theory. Every concept is introduced through a story, explained with clear mental models, and followed by tested Python code you can run yourself. You also get a free companion download with the sample Northwind dataset, scripts, exercises, and supporting files. You don’t need a GPU, cloud account, or a big budget. All you need is Python, curiosity, and a willingness to understand what your AI system is actually doing. By the end of the book, you won’t just know what RAG means. You’ll have built the complete “night shift” that turns an unruly document archive into knowledge your AI system can actually find.