Coding with AI

This is a short guide about working with AI as a software developer. It isn’t a list of prompts. It isn’t a productivity hack. It’s a way of thinking about what changes — and what doesn’t — when you have an AI collaborator available.

Who this is for

You’re a working developer. You’ve probably used an AI coding tool at least a few times — maybe Copilot, maybe ChatGPT, maybe something your team adopted. You’ve seen it do impressive things and you’ve seen it confidently produce garbage. You’re not sure where it actually fits in your work.

That uncertainty is reasonable. Most of what’s written about AI and coding is either breathless hype or existential dread. Neither is useful when you’re trying to ship software.

What this isn’t

This is not “50 ways to 10x your productivity.” That framing treats AI as a force multiplier for typing, which is the least interesting thing about it.

If the main thing you’re doing with AI is accepting autocomplete suggestions, you’re using a telescope as a doorstop. It works, technically. But you’re missing the point.

What this is

A few ideas — developed in practice, not theory — about how AI changes the work of writing software:

• Beyond Autocomplete — Why dialogue is more valuable than code generation, and what that means for how you work.

• AI as a Learning Partner — Concrete techniques for using AI to build understanding, not just output. This is the most underused capability.

• What You Bring to the Table — The human skills that matter more, not less, when AI is in the room.

A note on honesty

These ideas were developed in conversation with AI. Several of the examples come directly from building the platform you’re reading this on. It would be a strange kind of dishonesty to write about AI-assisted thinking while hiding that we used it.