How does Spring AI integrate with Spring Boot auto-configuration?

Spring AI follows standard Spring Boot auto-configuration conventions, which means zero boilerplate for the common case. When you add a provider starter to your dependencies and supply the required properties, Spring Boot auto-configures the AI beans you need without any @Configuration classes on your part.

Each provider starter (e.g. spring-ai-openai-spring-boot-starter) ships a spring/org.springframework.boot.autoconfigure.AutoConfiguration.imports file that registers its auto-configuration classes. Those classes use @ConditionalOnProperty and @ConditionalOnMissingBean guards so they activate only when needed and back off when you declare your own bean.

What gets auto-configured per provider:

• A ChatModel bean (e.g. OpenAiChatModel).
• An EmbeddingModel bean if the provider supports embeddings.
• A ChatClient.Builder prototype bean that injects the auto-configured ChatModel.
• Provider-specific @ConfigurationProperties bindings (API keys, base URLs, default model names, timeouts).

Minimal application.properties for OpenAI:

spring.ai.openai.api-key=${OPENAI_API_KEY}
spring.ai.openai.chat.options.model=gpt-4o
spring.ai.openai.chat.options.temperature=0.7

If you need to customise the HTTP client, add observability, or wire a custom RetryTemplate, declare a @Bean of the required type and Spring Boot's @ConditionalOnMissingBean will skip the auto-configured default in favour of yours.