Spring AI Masterclass for Generative & Agentic AI Developers

Learn to build a Spring Boot app that connects to OpenAI's LLM, fetches a dad joke, and exposes a REST API mimicking a Chat GPT clone that returns Simpsons trivia.

Create a Spring Boot project with Spring AI, connect to OpenAI via a REST API, configure spring.ai.openai.api-key in application.properties for a standalone app, while avoiding hard coded keys.

Learn to wire the spring ai chat client builder for OpenAI within a standalone spring boot app, using command line runner, autowiring, and an OpenAI starter to enable generative AI.

Learn to initialize a chat client via the builder, send a prompt to the OpenAI LLM, and print the response in a Spring Boot REST app for Simpsons trivia.

Build a RESTful API with Spring AI support by wiring a chat client in a REST controller, configuring system and user messages for a Simpsons trivia chat bot.

Create rest endpoint at /trivia with a get mapping in a Spring Boot app that returns a string from a prompt parameter and test locally on localhost:8080 for Simpsons trivia.

Set a system message and a user message to tailor how the LLM answers, enforcing Simpsons trivia focus and building the full prompt for a Spring Boot REST API query.

Learn best practices for managing OpenAI, Gemini, Grok, and Anthropic API keys by using environment variables, avoiding hard coded keys, and securing credentials in deployment.

Enable debug logging in spring to display chat completion requests, including system prompts and user prompts, revealing the JSON payload sent to the server for troubleshooting, debugging, and optimization.

Build a chat gpt UI with GitHub Copilot that uses /trivia endpoint to answer Simpsons questions. The page uses Simpsons colors, saves past questions in local storage, and adds celebrations.

Design an ai slop generator project that uses a large language model to rewrite tweets via templatized prompts, with a web front end and rest api.

Launch a spring ai starter project to build a slop generator with openai integration, exposing a restful api that generates and returns tweets.

Configure a Spring Boot REST API with a YAML-based OpenAI API key, and scaffold a Copilot-generated REST controller that transforms tweets via prompt templates.

Learn to use Spring AI's prompt template and chat client builder to generate five rewritten tweets from an original, using system prompts, user prompts, and modernization and emoji level controls.

Configure the llm by setting chat options in Spring AI, selecting GPT-5.2 and adjusting temperature to 0.99 and top_p to 0.95. Build the chat client with these options as defaults.

Understand how temperature and top p shape llm word choices, expanding or tightening the center circle, while top k may be unsupported by some models.

Improve prompts by moving the system prompt to a static final variable and refining the tweet rewrite engine to enforce five emoji-rich tweets in five styles.

Master structured llm outputs with Spring AI's bean output converter, mapping json to java records like tweet variant and new tweets, guided by a json schema.

Build the front end for the slop generator app using an index.html in the static folder. Enable local storage history, emoji and modernization controls, and LLM indicators during prompt processing.

Discover how retrieval augmented generation lets large language models access private data by retrieving relevant sources, and see how to implement it with Spring AI for answering questions.

Learn how retrieval augmented generation works by querying an llm with relevant pages from a book, retrieved from a vector store, and returning an answer augmented by this data.

Build a retrieval augmented generation app with a vector store by converting a PDF to searchable vectors and querying via a Spring Boot rest /ask endpoint powered by OpenAI.

Build a REST controller that uses a Spring AI chat client and embedding models to query a vector store, ingest content, and enable search via embeddings with dependency injection.

Load a book from a URL, convert each page into embeddings with OpenAI, and store them in a vector store to enable semantic search via a RESTful get endpoint.

Explore a restful RAG workflow that queries a vector store with OpenAI embeddings and retrieves five pages. Build augmented context and prompt the LLM to generate the answer.

Run the book rag application on localhost 8080, verify rag ingestion and the prompt, then test answering questions using top k pages from the book.

Build a rag web UI with GitHub Copilot by designing a Simpsons-inspired frontend, implement embeddings and a vector store for retrieval augmented generation, and add guardrails.

Explore how agentic applications gain memory by maintaining conversation history and context, and compare basic API memory solutions with a memory-enabled advisor approach using Spring AI.

Implement memory for a Spring AI chatbot using a rag controller and chat memory, assigning a unique CID per user and recording user and assistant messages to form history.

Reuse spring ai advisors to automatically manage memory with the memory chat advisor, attach it to the chat client, and implement memory-aware prompts for conversations.

Update the chat bot web ui to support spring ai's chat memory, implement a unique, locally stored cid, and adapt the rest api call to getMappingAsk for debugging.

Discover how MCP unlocks external tools for large language models, enabling real-time data, web access, file system actions, Jira, and social media posting within Spring AI code.

Build a Spring AI MCP server to expose model context protocol tools to an llm, using @McpTool and a MCP toolkit, then trigger the safe word through a RESTful endpoint.

Build an MCP client in a standalone Spring Boot app that connects to an MCP server, uses a tool callback provider, and integrates with OpenAI.

Create a Java MCP tool to save LLM responses to the local file system, generating a unique file name and writing text to a repos/text_responses directory.

Call MCP tools from a Spring client, save responses, and wire prompts into an agentic workflow. Learn to integrate tools and run natural language interactions to save outputs.

Learn to test, log, and observe generative and agentic AI applications, addressing model drift and the indeterministic nature of some LLMs, with metrics like latency and token usage.

Test MCP servers with the inspector, using tools like safe word and save the response. Connect to localhost:8085 via SSE, inspect JSON, and log results to disk.

Explore drift in AI testing and how to use one LLM as a judge to evaluate another, with spring boot’s framework for evaluations and fact checking.

Detect drift in LLMs with JUnit and Spring AI's FactCheckingEvaluator, using a test-driven setup to compare generator and judge outputs. Reveal evaluation results and surface drift.

Enable debugging and logging for Spring AI by configuring application.properties or application.yaml to print error logs and observe prompts, including system, user, and server interactions at the debug level.

Enable spring boot actuators in your restful app to gain health, info, and Prometheus metrics for development monitoring with micrometer, accessible at localhost:8080/actuator.

Extract a French menu via optical character recognition, translate to English, structure it as json, read aloud with text-to-speech, and generate images; store allergies and preferences in a vector store.

Configure text-to-speech attributes in application.properties for a spring boot app. Use ocr translation and voice-to-text to convert a menu image to json and mp3 with comet at speed 1.2.

Decorate the Spring app with a REST controller, initialize chat, image, and text-to-speech models, and wire them via a builder-based constructor before loading an image for OCR.

Learn to implement optical character recognition in a Spring Boot app using OpenAI. Build a get endpoint, load an image, and extract raw text with LLM prompts.

Translate OCR text for a menu to English using a system prompt, preserve HTML formatting, and return the result to be structured as JSON for processing in Spring AI app.

Transform translations into structured JSON by defining menu and menu item data structures, wiring an LLM-powered convertToJSON endpoint, and illustrating prompt-driven conversion and agentic workflow.

Create a RESTful endpoint that converts a menu to a prompt, calls the DALL·E image model at 1024 by 1792, and returns HTML with the generated image.

Convert a json menu into speech using an llm to generate mp3 audio via a restful api. Inject and configure the text-to-speech model with Spring Boot and OpenAI options.