Seeking feedback

[langchain4j]

Hi @michaelisvy, sorry for pinging you here - I couldn’t find another way to reach out. I hope you don’t mind.

I came across https://speakerdeck.com/michaelisvy/spring-ai-vs-langchain4j?slide=20 and wanted to get some feedback, especially on the "Support for Entities" category. Could you please share what problems you encountered there?

Thank you!

[langchain4j]

Regarding https://speakerdeck.com/michaelisvy/spring-ai-vs-langchain4j?slide=37: did you consider using high-level AI Services? They integrate nicely into Spring Boot application, here are few examples:

• https://github.com/langchain4j/langchain4j-examples/tree/main/spring-boot-example/src/main/java/dev/langchain4j/example/aiservice
• https://github.com/langchain4j/langchain4j-examples/tree/main/customer-support-agent-example/src/main/java/dev/langchain4j/example

[michaelisvy]

hi @langchain4j , sure! Nice to be in touch. Regarding entities, I was referring to this: https://speakerdeck.com/michaelisvy/spring-ai-vs-langchain4j?slide=32 I know it's also possible to do Structured Outputs in LangChain4J as shown here, but I find it complicated.

Hope that helps :)

[langchain4j]

@michaelisvy thank you for taking the time to provide feedback, I really appreciate! Would you mind elaborating a bit further? I'd love to better understand which specific aspects you found complicated.

There are currently 2 ways to implement this in LC4J:

Using AI Services

langchain4j.open-ai.chat-model.api-key=${OPENAI_API_KEY}
langchain4j.open-ai.chat-model.model-name=gpt-4o-mini

@AiService
public interface ActorFilmsGenerator {

    ActorFilms generate(String query);
}

@RestController
public class ActorFilmsController {

    private final ActorFilmsGenerator actorFilmsGenerator;

    public ActorFilmsController(ActorFilmsGenerator actorFilmsGenerator) {
        this.actorFilmsGenerator = actorFilmsGenerator;
    }

    @GetMapping("/generateActorFilms")
    public ActorFilms generateActorFilms() {
        return actorFilmsGenerator.generate("Generate the 10 most popular movies starring Bruce Willis");
    }
}

Using low-level API

langchain4j.open-ai.chat-model.api-key=${OPENAI_API_KEY}
langchain4j.open-ai.chat-model.model-name=gpt-4o-mini

@RestController
public class ActorFilmsController {

    private final ChatLanguageModel chatModel;

    public ActorFilmsController(ChatLanguageModel chatModel) {
        this.chatModel = chatModel;
    }

    @GetMapping("/generateActorFilms")
    public String generateActorFilms() {
        ChatRequest chatRequest = ChatRequest.builder()
                .messages(UserMessage.from("Generate the 10 most popular movies starring Bruce Willis"))
                .responseFormat(ResponseFormat.builder()
                        .type(JSON)
                        .jsonSchema(JsonSchemas.jsonSchemaFrom(ActorFilms.class).get())
                        .build())
                .build();
        ChatResponse chatResponse = chatModel.chat(chatRequest);
        return chatResponse.aiMessage().text();
    }
}

[michaelisvy]

understood thanks! I had missed it from the documentation, I'll add it to my sample application.

By the way, one small other thing which I thought was missing: I didn't find a way to load an image from th classpath. See here: https://speakerdeck.com/michaelisvy/spring-ai-vs-langchain4j?slide=33

Cheers, Michael.

[langchain4j]

@michaelisvy sure, this can be done as well:

@Service
class ImageService {

    @Value("classpath:images/scientist.jpg")
    private Resource imageResourceScientist;

    private final ChatLanguageModel chatModel;

    public ImageService(ChatLanguageModel chatModel) {
        this.chatModel = chatModel;
    }

    public String describeScientist() throws IOException {
        UserMessage userMessage = UserMessage.from(
                TextContent.from("can you describe this person? And what is written on top of his head?"),
                ImageContent.from(encodeInBase64(imageResourceScientist), "image/jpg")
        );
        ChatResponse chatResponse = chatModel.chat(ChatRequest.builder().messages(userMessage).build());
        return chatResponse.aiMessage().text();
    }

    private String encodeInBase64(Resource resource) throws IOException {
        return Base64.getEncoder().encodeToString(resource.getContentAsByteArray());
    }
}

Thank you for pointing this out, I will add more examples to the documentation!