From Assistants to Agents: Self-Improving Agentic Systems with Spring AI by Christian Tzolov

This talk discusses the principles of context window curation in AI agents, emphasizing the integration of learning and autonomous action to achieve better outcomes. The speaker illustrates how Spring AI provides a robust framework for building these agents, focusing on components like the ChatClient, Advisors, and Tool Calling APIs. Key topics covered include prompt engineering, conversation memory, and the use of various agentic patterns such as Recursive Advisors and Agent Skills for enhanced interaction and modular functionalities. Through live demonstrations using Spring AI's fluent API, participants gain insights into implementing generative AI in Spring applications, transforming complex agent behavior into simplified and maintainable code.