Spring AI: A Plain-English Guide for Java Developers

Use AI models in Java the way you already use databases. Spring AI gives you one API for chat models, embeddings, vector stores and tools, and swaps the provider with a property. This book teaches it by building, not by listing features: thirteen chapters, seventeen complete Spring Boot projects, and every program run on a free local model and on a paid cloud model before a word was written about it. What you build - A support desk that streams its replies, and a router that falls back to a local model when the cloud call fails- An invoice parser that returns Java records, with the numbers checked in plain Java- A duplicate-ticket detector using embeddings and cosine similarity, with no vector database at all- Semantic search over a product catalogue on PostgreSQL with pgvector- Question answering over your own PDFs, including why chunk size decides answer quality- An order assistant that queries a database and calls a live web API, with an approval gate- An MCP server that offers your application's tools to other AI applications- A four-layer test suite for answers that change on every run- Token, cost and latency accounting per feature, and a production checklist Written against a verified stack >Also included - Appendix A: 113 gotchas by topic, with the exact error text, so you can search for the message on your screen- An index, and a complete Maven project per chapter that opens in IntelliJ and runs with nothing but an API key, or nothing at all on a local model Who it is for Java and Spring Boot developers. No machine-learning background needed.