AI Agents Meet Adult Supervision

AI can generate thousands of lines of Java before your build finishes. The harder question is whether any of it belongs in production. This meeting digs into AI-generated Jakarta EE 11 applications, agent orchestration, MCP, tools, skills, portability, framework lock-in, and long-term maintainability. If “vibe coding” is coming for enterprise Java, standards may be the only thing standing between acceleration and chaos.

Sign up at https://www.javasig.com

Gerrit Grunwald

AI code generation is changing everything, and not always for the better. Every developer now has a tireless coding assistant that can produce hundreds of lines of Java in seconds. But code is cheap, software isn’t. AI generates what you describe, and if your description is vague, ad hoc, or tied to a specific framework version, the result will be too. In the enterprise, code written today must still be maintainable in five, ten, or fifteen years…by people who may not have been part of writing it. Framework lock-in, abandoned libraries, and undocumented conventions are already hard enough to manage without AI accelerating the problem. But when you give AI a specification-based contract like Jakarta EE, the generated code is portable, standardised, and guaranteed to have a support path for the foreseeable future. We will build a complete Jakarta EE 11 application, live, one spec at a time, entirely with AI assistance, to prove the point.

Ivar Grimstad

Building production-grade AI requires more than a model. It requires an ecosystem. The Java Community provides this ecosystem for the modern enterprise. Paired with Jakarta EE, it creates a formidable platform for AI Augmentation.
Come to this session to learn how to orchestrate enterprise-ready AI agents using Jakarta EE technologies in combination with AI augmentation approaches, such as tools, skills, and the Model Context Protocol (MCP).