Unlocking the New Era of AI Development: Exploring Spring AI Core Classes

Message (org.springframework.ai.chat.messages.Message)

Message represents the content sent to a large language model. Four MessageType variants exist: UserMessage (user input), SystemMessage (system instructions), AssistantMessage (model’s previous reply), and ToolMessage (tool output).

UserMessage userMessage = new UserMessage(message);
SystemMessage systemMessage = new SystemMessage(sysMsg);

Prompt (org.springframework.ai.chat.prompt.Prompt)

Prompt encapsulates a complete model input and can contain multiple Message objects and optional ChatOptions such as temperature.

Prompt prompt = new Prompt(userMessage, systemMessage);

ChatModel (org.springframework.ai.chat.model.ChatModel)

All conversational LLMs are abstracted by the ChatModel interface, which defines two primary methods: call – synchronous interaction. stream – streaming interaction.

The concrete implementation OpenAiChatModel is injected into the Spring container to handle communication with the underlying model.

Demo REST controller

Example controller wires the components and performs a chat call:

@RestController
public class TestController {
    @Resource
    private OpenAiChatModel chatModel;

    @GetMapping(value = "/deepseek/chat/call2")
    public String call2(@RequestParam(value = "message") String message,
                       @RequestParam(value = "sysMsg") String sysMsg) {
        UserMessage userMessage = new UserMessage(message);
        SystemMessage systemMessage = new SystemMessage(sysMsg);
        Prompt prompt = new Prompt(userMessage, systemMessage);
        ChatResponse chatResponse = this.chatModel.call(prompt);
        return Optional.ofNullable(chatResponse)
            .map(ChatResponse::getResult)
            .map(Generation::getOutput)
            .map(AssistantMessage::getText)
            .orElse("");
    }
}

Repository: https://github.com/Why1214/spring-demo.git (branch: spring-ai)