Spring AI with OpenAI and Milvus in Action

The AI-based service to make recommendations based on user documents

Overview

The application is designed as a service with no user interface.

Once the application is started, it will automatically load documents from the classpath (see application.yml) into the Milvus vector database. Then you can make POST requests to /ask/open-ai.

API documentation will be available once the application is started

MinIO (Multi-Cloud Object Store) will be available once the application is started. Login: minioadmin, Password: minioadmin

Technologies

• Java - version 21
• Maven - for building the application
• Spring Boot - version 3.4.3
• Spring AI - version 1.0.0-M6, model gpt-4-turbo
• Spring AI Tika - for extract metadata and text from documents
• Spring Boot Maven Plugin - for create Docker-Image
• Docker-Compose - infrastructure
• Milvus - vector database

Structure of the project

spring-ai-with-openai-and-milvus-in-action/
├── src/main/
|   ├── java/com/dudko/example/
|   |   ├── bootstrap/
|   |   ├── config/
|   |   ├── controller/             # controllers
|   |   ├── model/                  # service level of the domain, used in business logic and controllers
|   |   ├── service/                # business logic
|   ├── resources/
|   |   ├── templates/              # prompt and system message
|   |   ├── application.yml         # configs
├── compose.yml                     # docker-compose file
├── pom.xml                         # artifact of Maven
├── postman_collection.json         # collection of requests for Postman

How to try this project?

❗Don't forget to set the OPENAI_API_KEY environment variable before running the application.

docker-compose -f compose.yml up

Author:

Anatoly Dudko