LangChain4j Spring Boot

This project brings LangChain4j support in Spring Boot to build AI and LLM-powered applications. It provides integrations with LLM services and vector stores, as well as tools, chains, and AI services.

Using the starter projects in this repository, you gain the following advantages over using the vanilla LangChain4j libraries in Spring Boot:

• Autoconfiguration and unified configuration properties for models and vector stores
• HTTP infrastructure with RestClient, WebClient, and Jackson for all integrations
• Built-in observability with Micrometer, including LLM-specific metrics and traces
• Dev services with Docker Compose and Testcontainers for models and vector stores
• Service bindings for automatic connection configuration when running on Kubernetes.

🚀  Quick Start

Pre-Requisites

• Java 17+
• Docker/Podman
• Spring CLI

Getting Started

Using the Spring CLI, you can easily bootstrap a new Spring Boot application with LangChain4j support.

First, add the LangChain4j Spring Boot catalog providing the project templates.

spring project-catalog add langchain4j https://github.com/ThomasVitale/langchain4j-spring-boot

Then, create a new Spring Boot project for building an LLM Applications with LangChain4j and Ollama.

spring boot new myapp langchain4j-chat-ollama

Finally, navigate to the myapp folder and run the Spring Boot application. The first time you run it, it will take a while to download the Ollama container image used as a dev service based on the Testcontainers Spring Boot integration.

cd myapp
./mvnw spring-boot:run

You can now call the application that will use Ollama and llama3 to generate a text response. This example uses httpie to send HTTP requests.

http :8080/ai/chat message=="What is the capital of Italy?"

🦜 Models

OpenAI

Gradle:

implementation 'io.thomasvitale.langchain4j:langchain4j-openai-spring-boot-starter:0.9.0'

Configuration:

langchain4j:
  open-ai:
    client:
      api-key: ${OPENAI_API_KEY}

Example:

@RestController
class ChatController {
    private final ChatLanguageModel chatLanguageModel;

    ChatController(ChatLanguageModel chatLanguageModel) {
        this.chatLanguageModel = chatLanguageModel;
    }

    @GetMapping("/ai/chat")
    String chat(@RequestParam(defaultValue = "What did Gandalf say to the Balrog?") String message) {
        return chatLanguageModel.generate(message);
    }
}

Ollama

Gradle:

implementation 'io.thomasvitale.langchain4j:langchain4j-ollama-spring-boot-starter:0.9.0'

Configuration:

langchain4j:
  ollama:
    chat:
      model: llama3

Example:

@RestController
class ChatController {
    private final ChatLanguageModel chatLanguageModel;

    ChatController(ChatLanguageModel chatLanguageModel) {
        this.chatLanguageModel = chatLanguageModel;
    }

    @GetMapping("/ai/chat")
    String chat(@RequestParam(defaultValue = "What did Gandalf say to the Balrog?") String message) {
        return chatLanguageModel.generate(message);
    }
}

🫙 Vector Stores

Chroma

Gradle:

implementation 'io.thomasvitale.langchain4j:langchain4j-chroma-spring-boot-starter:0.9.0'

Example:

class ChromaDataIngestor {
    private final ChromaEmbeddingStore embeddingStore;
    private final EmbeddingModel embeddingModel;

    ChatController(ChromaEmbeddingStore embeddingStore, EmbeddingModel embeddingModel) {
        this.embeddingStore = embeddingStore;
        this.embeddingModel = embeddingModel;
    }

    public void ingest(List<Document> documents) {
        EmbeddingStoreIngestor ingestor = EmbeddingStoreIngestor.builder()
                .embeddingStore(embeddingStore)
                .embeddingModel(embeddingModel)
                .documentSplitter(recursive(300, 0))
                .build();
        ingestor.ingest(documents);
    }
}

Weaviate

Gradle:

implementation 'io.thomasvitale.langchain4j:langchain4j-weaviate-spring-boot-starter:0.9.0'

Example:

class WeaviateDataIngestor {
    private final WeaviateEmbeddingStore embeddingStore;
    private final EmbeddingModel embeddingModel;

    ChatController(WeaviateEmbeddingStore embeddingStore, EmbeddingModel embeddingModel) {
        this.embeddingStore = embeddingStore;
        this.embeddingModel = embeddingModel;
    }

    public void ingest(List<Document> documents) {
        EmbeddingStoreIngestor ingestor = EmbeddingStoreIngestor.builder()
                .embeddingStore(embeddingStore)
                .embeddingModel(embeddingModel)
                .documentSplitter(recursive(300, 0))
                .build();
        ingestor.ingest(documents);
    }
}

🌟 Examples

Check these examples to see LangChain4j and Spring Boot in action.

🛡️  Security

The security process for reporting vulnerabilities is described in SECURITY.md.

🖊️  License

This project is licensed under the Apache License 2.0. See LICENSE for more information.