AI tools are rapidly evolving from autocomplete assistants to autonomous agents capable of reasoning across entire systems. The promise isn’t just faster code, but leverage – where a single human decision can ripple outward into working software. This talk traces the recent epochs of AI in development and examines their impact and implications for software engineering practice. Drawing on Atomic Object’s experiences with AI in real projects, it outlines what it takes to make codebases and teams ready for AI collaboration, how to balance speed with ambition, and where model-driven thinking, whether via formal models or domain abstractions, offers especially useful leverage points. Rather than focusing on a single tool or technique, the session highlights patterns, risks, and opportunities across the current landscape, equipping attendees with a framework for understanding AI’s changing role in software development and practical ideas for preparing their own systems and teams.