Spring AI Example

This is a Spring Boot web application that uses Spring AI to talk with OpenAi models. It utilizes retrieval augmented generation to answer questions about the Iowa Hawkeyes football team for the 2023-24 season.

This repository was used as reference during a talk for the Central Iowa Java Users Group in January 2024 and Iowa Code Camp in October 2024. Slides for the talk can be found in this repository at Spring AI.pdf

Setup

• Create an OpenAI account and get an API key.
• Set up billing settings for your account with a small spending limit.
• Create an environment variable named SPRING_AI_OPENAI_API_KEY that is set to your OpenAI API key, or populate the key value in 'application.yaml`.
• Install Java 21 or later.
• Run ./gradlew bootRun to start the application.

Populating Vector Storage

• Once the application is running, you can populate the vector storage by running curl -X POST http://localhost:8080/stats/embeddings.
• This will create embeddings for all the data in src/main/resources/documents and store them in the vector storage file called vector-storage.json in your project's target/classes/vector-storage directory, so you only have to generate them once.

Using the API

• Once the vector storage is populated, you can use the API to get predictions from OpenAi using the embeddings in the vector storage.
• Run the following to get a response about the number of passing touchdowns.

curl --request POST \
  --url 'http://localhost:8080/stats' \
  --header 'Content-Type: application/json' \
  --data '{
     "question": "Who were the players with the most passing yards for Iowa in 2023?",
     "sport": "college football"
    }'

You'll receive a response like this:

{
  "teamName": "Iowa",
  "item": [
    {
      "statName": "Passing Yards",
      "year": 2023,
      "value": 1096.0,
      "playerName": "Deacon Hill"
    },
    {
      "statName": "Passing Yards",
      "year": 2023,
      "value": 505.0,
      "playerName": "Cade McNamara"
    }
  ]
}

Tech Stack

• Spring Boot 3.3
• Spring Boot Starter Web
• Spring AI
  • With Spring Boot Spring AI OpenAI Starter
• Java 21
• Lombok
• Gradle

Topics

• Tech Stack
• Simple Interactions
• Prompts and Templates
• Output Parsing
• RAG
• Function Calling