I'm not going to give you my full take on working with AI. I love it. I hate it. I have feelings.
What I do want to talk about is how I approach building for AI.
At my job I work on a design system and component library. A lot of that work is documentation. I genuinely like it. I care about teaching people. But it's obvious things are changing in our industry. For better or worse, I see plenty of both coming our way. In the end, it doesn't matter. We adapt.
I have always built and documented for the developer who wants to get better at their craft. I don't want to change that. There is, and always will be, value in understanding something just a little bit better. Going a little deeper. The good news is I don't think AI poses that big of problem. What I have learned is:
*What's good for a human is good for an agent*
They are large *language* models after all. Clear explanations, good structure, real examples. It's exactly what they're good at consuming.
Instead of turning my day job into writing SKILL.md files for a soulless agent to read, I take the human documentation and build two artifacts.
The goal hasn't changed: better outcomes for those leveraging the design system and component library.
If you are in the design systems business, AI doesn't replace good documentation. It makes it the foundation everything else is built on.