AI coding assistants made me slower — and that's the point

"Using AI to write better code more slowly" hit 1254 points. The thesis: AI tools make you faster at writing code but slower at understanding it. I've been thinking about this a lot because it describes exactly what happened to my coding workflow.

I use AI assistants daily. the truth is, claude Code handles a lot of my boilerplate — scaffolding projects, writing tests, refactoring code I wrote six months ago and no longer understand. But here's the thing: I don't ship the AI output directly. I read it, modify it, break it, fix it, and then commit it.

The "more slowly" part isn't a bug — it's a feature. AI writes code fast, but it writes generic code. It optimizes for the average case, not your specific case. The value I provide is knowing which parts to keep, which to rewrite, and which to delete entirely. That judgment comes from experience, not from the model.

My blog engine is a perfect example. The generator creates posts, the humanizer rewrites them to avoid AI detection, and I review the output. If I just published everything the AI generated, it would read like... well, like AI. The humanization step — shorter sentences, concrete examples, occasional swearing — that's the part that makes it worth reading.

Here's my actual workflow: I describe what I want in plain English. The AI writes a first draft. look, i rewrite the parts that sound generic. I add personal context — references to n8n, Coolify, trading bots, things I actually build. Then I publish. The whole process takes longer than pure AI generation, but the output is mine in a way that raw AI output never is.

The developers who will thrive in the AI era aren't the ones who write the fastest code. They're the ones who know what good code looks like and can guide AI toward it. Speed is cheap. Judgment is expensive.

. Curious what others think.