AI

March 6, 2026

In my last post I mourned the old ways and feared the future. This was in February 2025, so quite a long time before the November 2025 inflection point.

My main criticism at that time was, that AI just did not work for me, that it changed the way I write code. In essence, I didn’t have fun.

A lot changed since then. Since end of last year, I guess nearly all my code has been written by AI and I have effectively become a full-time reviewer, tester and product owner. And the best part? I’m having fun! Features that would take me days to implement, I can now do in a few minutes. Projects that I would have never even started, now appear realistic. Polish that I would have been too lazy to add now is just one prompt away.

To me this is a magical time.

A lot of previous knowledge regarding software development practice is now outdated. - How important is DRY really, when AI can generate boilerplate effortlessly? - Can there really be too many tests (given they execute fast enough)? - How do established team practices work, when the thing that used to be expensive (= code) becomes cheap?

As anybody, I’m trying to navigate this time somehow. We’ll see how everything works out, when the dust settles.