Right now, when working with AI-generated code, I’m trying more and more to focus on architectural thinking: what if what comes next is a small bug that I or someone else needs to fix?
Small bugs don’t need an entire app flow to be regenerated; it’s a small change, a quick PR, and off we go.
That’s the benefit of a strong architecture – you make the change where you need to make it and not all over the place.
If you have a framework for building services that works, AI inherits that infrastructure and leverages it, expanding it and making your code stronger and better.
If you’re rebuilding apps, front-facing, client-side, heavy-use apps- then you’re building recursive debt into your flow each and every time you do something.
The goal in leveraging AI is to think beyond the initial release – the first patch, the next feature, the follow-up branch – tackling those issues first is what ensures you minimize the recursive debt you might inadvertently adopt.