Spring AI RAG Demo

A Spring Boot demo showcasing Retrieval-Augmented Generation (RAG) using Spring AI and OpenAI GPT models. This project enables intelligent querying of local PDF documents by combining the power of LLMs with semantic search.

---

Features

• Ingest and vectorize PDF documents
• Perform semantic search with Spring AI
• Use GPT models with document context (RAG)
• Expose an API endpoint for intelligent querying

---

Requirements

• Java 23
• Maven
• Docker Desktop (for PostgreSQL with PGVector)
• OpenAI API Key

---

Dependencies

<dependencies>
    <dependency>
        <groupId>org.springframework.boot</groupId>
        <artifactId>spring-boot-starter-web</artifactId>
    </dependency>
    <dependency>
        <groupId>org.springframework.ai</groupId>
        <artifactId>spring-ai-openai-spring-boot-starter</artifactId>
    </dependency>
    <dependency>
        <groupId>org.springframework.ai</groupId>
        <artifactId>spring-ai-pdf-document-reader</artifactId>
    </dependency>
    <dependency>
        <groupId>org.springframework.ai</groupId>
        <artifactId>spring-ai-pgvector-store-spring-boot-starter</artifactId>
    </dependency>
</dependencies>

---

Getting Started

1. Configure Environment Variables

OPENAI_API_KEY=your_api_key_here

2. Update application.properties

spring.ai.openai.api-key=${OPENAI_API_KEY}
spring.ai.openai.chat.model=gpt-4
spring.ai.vectorstore.pgvector.initialize-schema=true

3. Add PDF Documents

Place your documents in:

src/main/resources/docs/

---

Run the Application

1. Start Docker Desktop (for PostgreSQL + PGVector).
2. Launch the Spring Boot app:

./mvnw spring-boot:run

---

Key Components

IngestionService

Handles PDF ingestion and vector store population:

@Component
public class IngestionService implements CommandLineRunner {

    private final VectorStore vectorStore;

    @Value("classpath:/docs/your-document.pdf")
    private Resource marketPDF;

    @Override
    public void run(String... args) {
        var pdfReader = new ParagraphPdfDocumentReader(marketPDF);
        TextSplitter textSplitter = new TokenTextSplitter();
        vectorStore.accept(textSplitter.apply(pdfReader.get()));
    }
}

---

ChatController

REST controller for document-aware queries:

@RestController
public class ChatController {

    private final ChatClient chatClient;

    public ChatController(ChatClient.Builder builder, VectorStore vectorStore) {
        this.chatClient = builder
                .defaultAdvisors(new QuestionAnswerAdvisor(vectorStore))
                .build();
    }

    @GetMapping("/")
    public String chat() {
        return chatClient.prompt()
                .user("Your question here")
                .call()
                .content();
    }
}

---

Making Requests

Send a query:

curl http://localhost:8080/

Response will include GPT-generated content enhanced with document context.

---

Architecture

Component	Description
Document Ingestion	Reads & splits PDF documents
Vector Store	Stores embeddings using PGVector
Semantic Search	Retrieves relevant document chunks
GPT Integration	Generates responses with embedded document knowledge

---

Best Practices

Document Ingestion

• Avoid reinitializing vector store unnecessarily
• Schedule periodic ingestion for updated documents

Query Optimization

• Monitor API token usage
• Cache frequent queries
• Rate-limit endpoint to avoid misuse

Security

• Secure API endpoints with authentication
• Encrypt and protect document content
• Store API keys securely (avoid hardcoding)

---