Skip to main content

Spring AI: Working with Google Vertex AI (Gemini)

Greetings!

Over the last few articles, we've learned the fundamentals of Spring AI using different concepts like prompts, memory, RAG, tools, and agents. In this article, we'll focus on something much simpler—connecting a Spring Boot application to Google Vertex AI and chatting with Gemini.

If you've already used the OpenAI integration, you'll notice that the application code is almost identical. The biggest difference is authentication, and that's exactly what we'll cover.

Why Vertex AI?

Google provides two ways to access Gemini models.

  • Google AI Studio - Great for learning and quick experiments using an API key.
  • Google Vertex AI - Enterprise platform for production workloads.

Vertex AI offers:

  • Enterprise authentication using Google Cloud IAM
  • Security and access control
  • Monitoring and logging
  • Quotas and billing management
  • Integration with other Google Cloud services

For production applications, Vertex AI is the recommended choice.

Local Vertex AI Setup

Unlike OpenAI, Vertex AI doesn't normally use an API key. Instead, it authenticates using Application Default Credentials (ADC). Spring AI automatically discovers these credentials and uses them whenever it communicates with Vertex AI.

1. Install Google Cloud CLI

Download and install the Google Cloud CLI.

https://cloud.google.com/sdk

Verify the installation.

bash
gcloud version

2. Login

bash
gcloud auth login

Your browser opens and asks you to sign in with your Google account.

3. Select your project

bash
gcloud config set project YOUR_PROJECT_ID

Example

bash
gcloud config set project spring-ai-demo

Verify it.

bash
gcloud config list

4. Create Application Default Credentials

This is the important step.

bash
gcloud auth application-default login

Google stores your credentials locally. Spring AI automatically discovers these credentials and obtains access tokens whenever it calls Vertex AI. You don't need to write any authentication code.

5. Verify everything

bash
gcloud auth application-default print-access-token

If you receive an access token, your machine is correctly configured. You're now ready to build the application.

Create the Spring Boot Project

Generate a standard Spring Boot project with:

  • Spring Web

Then add the Spring AI dependency org.springframework.ai:spring-ai-starter-model-google-genai.

application.yml

yaml
spring:
  ai:
    google:
      genai:
        project-id: your-project-id
        location: us-central1

        chat:
          model: gemini-2.5-flash
          temperature: 0.7

Notice something interesting. There is no API key. Spring AI automatically uses the Application Default Credentials that you created earlier.

Chat Service

java
package com.slmanju.demo;

import org.springframework.ai.chat.client.ChatClient;
import org.springframework.stereotype.Service;

@Service
public class GeminiService {

    private final ChatClient chatClient;

    public GeminiService(ChatClient.Builder builder) {
        this.chatClient = builder.build();
    }

    public String chat(String message) {
        return chatClient.prompt(message)
                .call()
                .content();
    }
}

As you can see, the code is exactly the same as our previous OpenAI examples. Changing AI providers usually requires changing only the dependency and configuration.

REST Controller

java
package com.slmanju.demo;

import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RequestParam;
import org.springframework.web.bind.annotation.RestController;

@RestController
@RequestMapping("/chat")
public class ChatController {

    private final GeminiService service;

    public ChatController(GeminiService service) {
        this.service = service;
    }

    @GetMapping
    public String chat(@RequestParam String message) {
        return service.chat(message);
    }
}

Run the Application

Start the Spring Boot application and call:

text
GET http://localhost:8080/chat?message=Explain dependency injection in Spring.

Or using curl:

bash
curl "http://localhost:8080/chat?message=Explain%20dependency%20injection%20in%20Spring"

You should receive a response generated by Gemini.

Summary

Connecting Spring AI to Google Vertex AI is surprisingly straightforward.

The application code is almost identical to the OpenAI integration. The main difference is authentication, where Vertex AI uses Application Default Credentials instead of an API key.

Once ADC is configured, Spring AI automatically authenticates with Vertex AI, allowing you to focus entirely on building your AI application.


Comments

Post a Comment