The Kind of Training That ACTUALLY WORKS For Agentic AI

Agentic AI is changing how software is built, but most AI training for software developers doesn’t actually work. In this video, Emily Bache explains what agentic AI really is, why common approaches like prompt lists and toy examples fail, and what software engineers should focus on instead. As AI agents become part of everyday software development, developers are being told to “just experiment” or take generic agentic AI courses that don’t transfer to real production code. Drawing on years of experience in modern software engineering, Emily Bache breaks down how to learn agentic AI in a way that improves real engineering outcomes, not just short-term productivity. This video covers: • What agentic AI and AI agents actually are, and how they differ from simple coding assistants • Why most agentic AI training, courses, and tutorials don’t translate to real software development • The limits of exercises like FizzBuzz and toy coding problems when learning coding with AI • How core softwar