Spring AI Vector Store with Infinispan

What You Will Learn

How to use Infinispan as a Spring AI vector store to store document embeddings and perform semantic similarity search. The auto-configuration handles schema registration and cache creation automatically.

Prerequisites

Java 17+
Docker or Podman (optional)

You can start an Infinispan Server manually with Docker or Podman:

docker run -it --rm -p 11222:11222 -e USER=admin -e PASS=password quay.io/infinispan/server:latest

Tip	You can replace `docker` with `podman` in the command above if you use Podman.

Tip	If no server is running on `localhost:11222`, the application automatically starts an Infinispan container using Testcontainers.

Step 1: Add Dependencies

Add the Infinispan Spring AI vector store auto-configuration and a local embedding model:

<dependency>
   <groupId>org.infinispan</groupId>
   <artifactId>spring-ai-autoconfigure-vector-store-infinispan</artifactId>
</dependency>
<dependency>
   <groupId>org.infinispan</groupId>
   <artifactId>spring-ai-infinispan-store</artifactId>
</dependency>
<dependency>
   <groupId>org.springframework.ai</groupId>
   <artifactId>spring-ai-transformers</artifactId>
</dependency>

The TransformersEmbeddingModel runs locally using ONNX. No API keys or external services needed.

Step 2: Configure the Connection

In application.properties, configure the Infinispan connection:

infinispan.remote.server-list=127.0.0.1:11222
infinispan.remote.auth-username=admin
infinispan.remote.auth-password=password

The vector store auto-configuration creates the cache and registers the Protobuf schema automatically.

Step 3: Provide an Embedding Model

Define an EmbeddingModel bean. The auto-configuration uses it to convert text to vectors:

@Configuration(proxyBeanMethods = false)
public class Config {

   @Bean
   public EmbeddingModel embeddingModel() {
      return new TransformersEmbeddingModel();
   }
}

Step 4: Store Documents

Add documents with metadata to the vector store. Each document is automatically embedded and stored in Infinispan:

   @PostMapping(path = "/load")
   public String loadData() {
      List<Document> documents = List.of(
            new Document("Infinispan is a distributed in-memory key/value data store",
                  Map.of("source", "docs", "topic", "overview")),
            new Document("Infinispan supports vector search for AI use cases",
                  Map.of("source", "docs", "topic", "ai")),
            new Document("Spring AI provides a unified API for AI engineering",
                  Map.of("source", "spring", "topic", "ai")),
            new Document("Infinispan can be used as a vector store with Spring AI",
                  Map.of("source", "tutorial", "topic", "integration")));

      vectorStore.add(documents);
      return "Loaded " + documents.size() + " documents";
   }

Step 5: Search by Semantic Similarity

Query with natural language. The vector store finds semantically similar documents:

   @GetMapping(path = "/search")
   public List<Document> search(@RequestParam(defaultValue = "How can I use Infinispan with AI?") String query) {
      return vectorStore.similaritySearch(
            SearchRequest.builder()
                  .query(query)
                  .topK(3)
                  .similarityThreshold(0)
                  .build());
   }

Step 6: Run the Tutorial

mvn spring-boot:run

Then load documents and search:

curl -X POST http://localhost:8080/load
curl "http://localhost:8080/search?query=How+can+I+use+Infinispan+with+AI"

Results are ranked by cosine similarity. Documents about vector search and Spring AI integration will rank highest.

What’s Next

Try Spring AI Chat Memory to persist chat conversations in Infinispan
Try LangChain4j Embedding Store for direct vector usage without Spring AI