A friend sent a screenshot that triggered me. It's all funny, of course. But not really. We've reached the point where building something is easier than figuring out why you should build it. Because now we're all vibe coders.
I'm in this trap myself. Surrounded by agents like pillows in a Turkish apartment. Claude Code here, Opus 4.6 there, GPT-Pro sprinkled on top. Every five minutes a "brilliant" idea flashes in my head, and instead of letting it die a natural death, I feed it to the terminal.
In the past six months I bought 10 domains. Resurrected a 2019 project that nobody needed even back then, simply because an MVP can be assembled in 15 minutes. The terminal and IDE became my main sources of cheap dopamine, replacing reels and (a little) sugar.
I used to be held back by constraints. Want a product? Build a team, find a budget, work through partner disagreements. All of that worked as a BS filter. If an idea survived three months of approvals, it at least had some merit.
Now for $200 a month you have a team that doesn't argue, doesn't sleep, and doesn't ask for equity. And you feel AI-omnipotent. But it's the omnipotence of a child given a real lightsaber. Sparks everywhere, looks beautiful, but half the apartment is already on fire.
My main insight: the most important skill right now isn't prompt engineering. It's the ability to look at a technically simple, cheap-to-build, and "cool" idea and say: no. I'm not going to do this.
We're on the brink of the slop-product era, and the internet will be flooded not just with bad texts but with entire working services that exist only because someone vibe-coded them.
Learning to deny myself the pleasure of creating. Not going well so far.