I don't really write code anymore. But almost every tool I use is something I built for myself.

Sounds contradictory, but it really isn't anymore. If you ask AI to build something these days, it usually can. The real problem is somewhere else — most of the time, I don't even know what to ask for.

At one point I thought, "I wish my editor's UI looked more polished." If I'd just told AI to build that, I probably would've gotten something flashy with features I'd never actually use. But when I asked myself why I wanted it polished, the answer was somewhere else entirely. I wanted to run my channel better. That meant fixing my analytics tools first, not the UI. As I dug into the analytics tools, the real problem turned out to be a JSON file I kept editing by hand—and occasionally breaking. Most of the work isn't building the solution. It's discovering what the problem actually is.

This blog is where I write that process down.

What I'm doing here:

  • Documenting real personal projects I actually use and keep improving
  • Showing the full arc from "I vaguely want this" to something that actually works
  • AI writes most of the code. My job is figuring out what I actually want, designing the workflow, and deciding what should be built first

What this isn't:

  • "Just use this one prompt" style showcases
  • Reviews or trend roundups of tools I haven't actually used

In a world where AI can build almost anything, I think the skill that actually matters isn't coding — it's knowing precisely what you want. This blog is a record of those attempts.