Spring AI for Enterprise Java — Examples

Small, runnable Spring AI 2.0.0 examples for enterprise Java / Spring Boot teams — one module per episode of the Spring AI for Enterprise Java video series.

Each module is a self-contained Spring Boot app you can run on its own. This is a learning companion, not a framework — no shared abstractions, no magic. Read the code, run it, adapt it.

• Watch: @TheStackUnderflow
• Read: thestackunderflow.com/tutorials

Requirements

Tool	Version	Why
JDK	21+	Spring AI 2.0.0 → Spring Boot 4.0 → Spring Framework 7.0 requires Java 21 to build.
Gradle	(wrapper included)	./gradlew — no local Gradle install needed.
API key	OpenAI (and/or Anthropic)	Set via env — see .env.example.

Quick start

git clone https://github.com/thestackunderflow/spring-ai-enterprise-examples.git
cd spring-ai-enterprise-examples

cp .env.example .env          # then put your real key in .env (git-ignored)
export OPENAI_API_KEY=sk-...   # or load .env however you like

# Run one episode's example:
./gradlew :01-what-is-spring-ai:bootRun

Never commit a real key. .env is git-ignored; the examples read keys from environment variables only. application.properties uses ${OPENAI_API_KEY:} so the app starts (for tests) even without one.

Modules

#	Module	Concept	Video
01	01-what-is-spring-ai	ChatClient in ~10 lines; the enterprise seam	(link on publish)
02	02-why-spring-boot-teams	DI, starters, auto-config — the patterns that transfer	soon
03	03-chatclient-explained	The fluent chain + reading token usage / metadata	soon
04	04-model-portability	One API, swappable providers (OpenAI · Anthropic · Ollama)	soon
05	05-structured-output	.entity() — AI text → typed Java records	soon
06	06-rag-vector-stores	RAG: docs → embeddings → vector store → grounded answer	soon
07	07-tool-calling	@Tool — the model calls your Java services	soon
08	08-mcp-integration	MCP client + server (the standard tool port)	soon
09	09-observability	Traces, metrics, token/latency for every model call	soon
10	10-evaluation-guardrails	RelevancyEvaluator — test AI output like code	soon
11	11-security-compliance	Secrets, boundaries, SafeGuardAdvisor, audit	soon
12	12-production-blueprint	Everything wired together, end to end	soon

Modules land as each episode ships; folders for upcoming ones hold a short README until then.

Versioning

Pinned to Spring AI 2.0.0 (see build.gradle.kts). When Spring AI releases a new version, examples are updated alongside a "What changed" video — subscribe to catch those.

License

Apache-2.0. Examples are original, written against the Spring AI 2.0.0 reference docs; the concepts are inspired by Ken Kousen's Apache-2.0 Spring AI Training Course.