Speaker Details

AI features might be easy to prototype, but as you add more features and dependencies, they can become bulky, slow to start, and hard to deploy. In this talk, we’ll build three compact examples — semantic search, booking, and automation — and focus on making them fast, predictable, and cheap to run in production.

For search, we’ll use Spring AI to turn natural-language queries into embeddings and retrieve relevant context for an LLM-backed online store database. For booking, we’ll use LangChain4j-based tool calling with a predefined set of actions in an application that emulates dinner reservations. For automation, we’ll build a workflow turning user input into social media artifacts and GitHub PRs.

We'll also explore the benefits of building such applications as GraalVM native images, and talk about integrations, best practices, and everything you need to know to build such applications with confidence.