Using AI as a design engineer

As a design engineer, a role that's about craft, thoughtfulness and creativity, I should probably be skeptical of AI. It's not particularly great at those things — at least for now. But despite that, I kinda enjoy using it.

Preface

When I work on something, I experiment a lot. A big part of my job is trial and error until something feels right.

AI helps by shortening that loop. Instead of spending hours on a concept I'm unsure about, I try it out in minutes and throw it away if it doesn't land.

There's an important distinction though: I don't use AI to come up with ideas or to replace my thinking. I use it to accelerate my workflow.

Setup

The first thing I do on a project is set up a small pile of rules for the codebase. You already have a strong sense of how you want certain things done — cn over template literals, design-system usage, animation conventions. Write those down once.

I keep rules short and to the point, and let them reference other files.

Models

I use Claude as my daily driver for most coding tasks, Codex for code review on PRs, and ChatGPT when I just want to ideate. For UI from Figma frames, the Figma MCP plus a good rule set has been surprisingly effective.

Conclusion

AI has changed my day-to-day. The upside is obvious: experiment more, ship faster. The downside is that it's very easy to outsource your thinking to it, and that shows up in the output.

Use AI for the tedious, repetitive parts. Keep the taste and the judgement yours.