AI is changing how we write software — but what about software architecture? Vibe Coding with LLMs enables quick results, but often creates emergent structures that are difficult to understand. In a world where systems can be built cheaply, architectural work isn't becoming less important—it's becoming essential: cycles are shorter, feedback is faster, and design and implementation are more tightly intertwined.

In this talk, I'll dive into how LLMs are reshaping architectural practices and how solid architectural principles can help you make sense of AI-generated code. Using concrete tools like AI review agents, LLM coding workflows, and custom analysis agents, I'll demonstrate how AI-supported projects can maintain quality and speed. I'll also discuss proven architectural methods that hold up in AI environments, including lightweight reviews, AI-supported reflection, and set-based design.

Participants will learn:
- How to turn chaotic "vibe coding" into a disciplined approach
- How to use LLMs for continuous architectural guidance
- How to integrate architectural practices into AI workflows

This session is for architects, tech leads, and developers who want to build professional software systems with AI, rather than just write cheap code.