Spring AI

Java teams get a first-class AI stack in Spring AI, the official Spring project that applies the ecosystem's portable-abstraction philosophy to model integration. A fluent ChatClient API, styled after WebClient and RestClient, talks to Anthropic, OpenAI, Amazon Bedrock, Google, Mistral, DeepSeek, Ollama, and other providers behind one interface, while the Advisors API packages recurring generative patterns such as retrieval augmentation and persistent chat memory into composable components. Coverage extends to structured output mapping onto POJOs, tool calling, observability, and vector store abstractions spanning more than 15 backends including PostgreSQL pgvector, Pinecone, and Weaviate, plus first-class Model Context Protocol support through dedicated client and server Boot starters. Projects pull it from Maven Central via Spring Boot starters at start.spring.io; the 1.1 line targets Spring Boot 3.5 while the 2.x line tracks Boot 4. Apache-2.0 licensed and maintained by the Spring team, it has become the default path for enterprise Java shops adding LLM features to existing applications.