Spring AI Tutorial: Configure Redis Vector Store & Build RAG with Advisors

Configure Redis Vector Database

Add the Spring AI Redis vector store starter dependency to your project:

<dependency>
  <groupId>org.springframework.ai</groupId>
  <artifactId>spring-ai-starter-vector-store-redis</artifactId>
</dependency>

Then configure Redis connection properties (host, port, password, index name, etc.) as shown in the accompanying screenshot.

Enable RAG with Vector Store Advisor

To implement retrieval-augmented generation, include the vector store advisor dependency:

<dependency>
  <groupId>org.springframework.ai</groupId>
  <artifactId>spring-ai-advisors-vector-store</artifactId>
</dependency>

Create a QuestionAnswerAdvisor using the builder pattern because the class is package-private in Spring AI 1.1.2 and cannot be instantiated directly:

The Spring AI starter auto-configures a VectorStore bean at startup; inject it directly into your components.

During runtime the author encountered an EmbeddingModel unavailable error and resolved it by manually defining an embedding model bean:

Ingest Documents via ETL Pipeline

Spring AI provides ETL interfaces; use a Reader to convert files into Document objects and write them to the vector store:

Test the RAG Conversation

Run a sample chat; the advisor retrieves relevant context from Redis and augments the prompt. The screenshot shows the input text and the model's answer:

Verify Stored Vectors in Redis

Inspect the Redis index to confirm documents and embeddings were persisted correctly: