AI coding tools can be incredibly useful, but they can also be confidently wrong, overly helpful, and occasionally creative in exactly the wrong places.

This talk is a practical handbook for developers who already use AI and want to use it better. Rather than focusing on clever prompts or shiny demos, we’ll look at ten simple rules for working with AI in real software development: how to give it the right context, make it investigate before editing, encode repeatable instructions, keep business knowledge discoverable, verify its work with deterministic tools, and avoid turning your codebase into one giant prompt.

The goal is not to make AI autonomous. It is to make AI useful, predictable, and easier to review.

Expect practical examples, patterns you can apply to an existing codebase, and a healthy amount of skepticism about anything that says “done” before the tests have run.