Your Code Isn’t AI-Friendly (Yet): Fixing the Complexity That Breaks AI Agents
AI cannot read between the lines. It struggles with missing intent, implicit concepts, and surprising dependencies. These aren't new problems, but much of our industry has fallen short in the past. That comes back to bite us.
Where a human developer will ask questions and seek clarification, an AI proceeds with what’s there, making its best guess from patterns in code that was never designed to be unambiguous.
In this talk, Adam Tornhill shows how to fix the system itself so agents stop guessing. All recommendations are grounded in AI research and cognitive psychology. You’ll learn:
The key principles behind AI-friendly codebases.
AI-driven refactoring patterns that make those principles concrete.
How to build software that both humans and machines can understand, modify, and evolve safely.
Zero fluff.