Enhance your Java Spring Applications with AI

I came across the release notes for Spring AI 0.8.0 and i thought i might give it a try and it’s frighteningly simple to do so. Here is a short sample Project: https://github.com/sijakubo/spring-ai-test

You just have to:

• include the Spring AI dependencies
• provide an AI token
• call the Spring AI proxy (e.g. ChatClient, ImageClient)

Supported models are:

Chat Models

• OpenAI
• Azure Open AI
• Amazon Bedrock
• Anthropic’s Claude
• Cohere’s Command
• AI21 Labs’ Jurassic-2
• Meta’s LLama 2
• Amazon’s Titan
• Google Vertex AI Palm - Gemini support coming soon (follow the WIP branch)
• HuggingFace - access thousands of models, including those from Meta such as Llama2
• Ollama - run AI models on your local machine

Text-to-image Models

• OpenAI with DALL-E
• StabilityAI

Embedding Models

• OpenAI
• Azure OpenAI
• Ollama
• ONNX
• PostgresML
• Bedrock Cohere
• Bedrock Titan
• Google VertexAI

Include Spring AI Dependencies:

just add the Maven BOM for the Spring AI to your gradle dependency management and include the actual Spring AI spring boot Starter dependency

build.gradle.kts

dependencyManagement {
    imports {
        mavenBom("org.springframework.ai:spring-ai-bom:0.8.0")
    }
}

dependencies {
    ...
    implementation("org.springframework.ai:spring-ai-openai-spring-boot-starter")
    ...
}

provide an AI token

In order to use an AI Service, you’ll have to provide an actual token. You can do so within the application.properties. You need to provide the token for the AI Service to use within your Application. E.g.:

spring.ai.openai.api-key=<OPEN_AI_API_KEY>
spring.ai.openai.image.api-key=<OPEN_AI_API_KEY>

spring.ai.azure.openai.api-key=<AZURE_API_KEY>

You also have the option to configure the usage of the designated AI Service. E.g. using the OpenAI Service, you can configure e.g. these options:

Call the Spring AI proxy

In order to actually use the AI Service, you can call the designated Proxy Client.

When just using the ChatClient you can prompt the AI questions, just as if you would chat with the AI. E.g.

String quote = chatClient.call("Tell a random funny 'The Office' quote");

or

String translatedText = chatClient.call("Translate the following text from langauge: %s to language: %s. The text is: %s"
    .formatted(
        translationResource.sourceLanguage(),
        translationResource.targetLanguage(),
        translationResource.text()
    ));

Conclusion

It’s utterly simple to integrate an AI Service into your application. As often, the only limit is your ability to identify useful UseCases for this. Spring provides easy methods to integrate several AI Services by abstracting beyond the actual service APIs which makes it easy to exchange between multiple Services and evaluate their results to your personal needs.