Skip to content

Latest commit

 

History

History
140 lines (103 loc) · 4.16 KB

README.md

File metadata and controls

140 lines (103 loc) · 4.16 KB

Spring AI Structured Output Demo

Transform AI responses into type-safe Java objects effortlessly with Spring AI's structured output capabilities. This project demonstrates how to convert natural language responses from Large Language Models (LLMs) into strongly-typed Java objects without writing complex parsing logic.

Why Structured Output Matters

Working with AI models typically means dealing with unstructured text responses. Converting these responses into usable data structures traditionally requires:

  • Writing custom parsing logic
  • Handling edge cases and malformed responses
  • Maintaining complex regex patterns or string manipulation code
  • Implementing error handling for parsing failures

Spring AI eliminates these challenges by providing built-in conversion capabilities that transform AI responses directly into Java objects. This project showcases this functionality through a practical example: converting information about technical authors into structured data.

Project Overview

This application demonstrates structured output conversion by:

  1. Taking user input for a technical author's name
  2. Querying an AI model for information about the author
  3. Automatically converting the response into a strongly-typed AuthorInfo object
  4. Displaying the structured information to the user

Requirements

  • Java 17 or higher
  • OpenAI API key
  • Spring Boot 3.2 or higher
  • Spring AI dependencies

Project Structure

src/
├── main/
│   ├── java/
│   │   └── io/spring/examples/
│   │       ├── Application.java       # Main application class
│   │       └── AuthorInfo.java       # Data structure for author information
│   └── resources/
│       └── application.properties    # Configuration properties

Key Features

Structured Data Model

The AuthorInfo record defines our structured output format:

public record AuthorInfo(
    String name,
    String primaryGenre,
    List<String> notableBooks,
    int yearsActive,
    String specialization
) {}

Automatic Conversion

Spring AI handles the conversion from AI response to Java object automatically:

var authorInfo = chat.prompt()
    .user(u -> {
        u.text("Please give me information about ${author}");
        u.param("author", authorName);
    })
    .call()
    .entity(AuthorInfo.class);  // Automatic conversion to AuthorInfo

Getting Started

  1. Set your OpenAI API key:

    export OPENAI_API_KEY=your-api-key-here

  2. Run the application:

    ./mvnw spring-boot:run

  3. Enter author names when prompted:

    Who is your favorite technical author? (e.g., 'Rod Johnson', or type 'exit' to quit)
USER: Rod Johnson

Response Format

The application displays structured information about the author:

=== Author Analysis ===
Name: Rod Johnson
Primary Genre: Technical Writing
Notable Books: Expert One-on-One J2EE Design and Development, Expert One-on-One J2EE Development without EJB
Years Active: 20
Specialization: Enterprise Java Development and Spring Framework

Technical Details

Configuration

The application uses Spring Boot's configuration properties:

spring.application.name=structured-output
spring.ai.openai.api-key=${OPENAI_API_KEY}
spring.ai.openai.chat.options.model=gpt-4

Error Handling

Spring AI's structured output converter handles various edge cases:

  • Malformed AI responses
  • Missing fields in the response
  • Type conversion errors

Benefits of Using Spring AI's Structured Output

  1. Type Safety: Get compile-time checking for your AI response structures
  2. Clean Code: Eliminate boilerplate parsing code
  3. Reliability: Robust handling of edge cases and malformed responses
  4. Maintainability: Changes to response structure only require updating the data model
  5. Integration: Seamless integration with Spring's ecosystem

Learn More

For more information about Spring AI and structured output conversion, visit:

Built with ❤️ using Spring AI