pgvector

Summary

This article describes how to use pgvector with Spring AI.

The complete source is available on GitHub JavaAIDev/pgvector-sample.

Spring AI supports different types of vector stores. pgvector is used as an example.

To use pgvector, Spring AI module spring-ai-pgvector-store must be added. The easiest approach is using Spring Boot starter for pgvector.

Spring Boot starter for pgvector

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

This Spring Boot starter will create required beans to use pgvector. We can simply provide configurations. The configuration prefix of pgvector is spring.ai.vectorstore.pgvector. initializeSchema is set to true to initialize database table schema required for Spring AI.

pgvector configuration

spring:
  application.name: pgvector
  ai:
    openai:
      api-key: ${OPENAI_API_KEY:demo}
      chat:
        options:
          model: gpt-4.1-mini
          temperature: 0.0
      embedding:
        options:
          model: text-embedding-3-small
    vectorstore:
      pgvector:
        tableName: vector_store
        schemaName: public
        schemaValidation: false
        initializeSchema: true
  datasource:
    url: jdbc:postgresql://localhost:5432/postgres
    username: postgres
    password: postgres

The application code can use VectorStore directly.