A support-triage AI agent in Java with Spring AI 2.0.1: one structured workflow step, a bounded tool loop, tenant data the model never sees, and refunds that wait for a human. Plus the guardrails Spring AI now ships and the ones you still write yourself.
A filtered list of MCP servers worth connecting to a coding assistant, with the specific flag that keeps each one from flooding your context window or writing to production. Plus the cases where a plain CLI beats an MCP server by 20x.
On a well-built Spring Boot app, exposing your product to AI agents is almost trivial: annotate a service method, add one starter. The wiring is the easy part. Choosing what to expose and securing it is the real work. Here is the honest version, on Spring AI 2.0.
RAG means handing your AI the right page from your own documents before it answers. It is powerful when your knowledge base is large, changing, or per-customer. It is also something a lot of teams build too early. Here is the honest decision.
You run a Java product and want to add an AI feature. There are two real framework choices. Here is the honest comparison across providers, RAG, tools, Spring fit, and the one thing that decides it for most teams.
The API price page says a few dollars per million tokens and looks cheap. That is not what your AI feature will cost. Here is the real formula, a worked example, the levers that cut the bill, and when self-hosting actually pays.
