Currently the only way to create an instance of an AI Service is to manually use the builder:

Assistant assistant = AiServices.builder(Assistant.class)
    .chatLanguageModel(model)
    .tools(tool)
    .chatMemory(MessageWindowChatMemory.withMaxMessages(10))
    ...
    .build();

While this is pretty simple and flexible, this can be made even more simple and natural for the Spring Boot users.

With this change, any interface annotated with @AiService will be automatically registered as a bean and configured to use all the LangChain4j components (beans) available in the context (e.g. chat model, chat memory, tools, etc.).

Here is an example of the simplest declarative AI Service:

@AiService
interface Assistant {

    String chat(String userMessage);
}

langchain4j.open-ai.chat-model.api-key=${OPENAI_API_KEY}

By configuring this property, an instance of the OpenAiChatModel will be autoconfigured as a bean. (This is not new, it was already possible with langchain4j-open-ai-spring-boot-starter.)

<dependency>
    <groupId>dev.langchain4j</groupId>
    <artifactId>langchain4j-spring-boot-starter</artifactId>
    <version>0.29.0</version>
</dependency>

<dependency>
    <groupId>dev.langchain4j</groupId>
    <artifactId>langchain4j-open-ai-spring-boot-starter</artifactId>
    <version>0.29.0</version>
</dependency>

When the application starts, autoconfiguration in langchain4j-spring-boot-starter will scan the classes of the project (starting from the package where @SpringBootApplication is located) and will find all interfaces annotated with @AiService.
It will use all available components (such as OpenAiChatModel in this case) when creating an instance of an AI Service.

It is also possible to customize which components should be used for the particular AI Service:

@Bean
ChatLanguageModel myCustomChatModel() {
    return OpenAiChatModel.withApiKey(...);
}

@AiService(wiringMode = EXPLICIT, chatModel = "myCustomChatModel")
interface Assistant {

    String chat(String userMessage);
}