vibe coding is boring

vibe coding is boring.

not because ai is boring. ai is moving insanely fast.

it started with ChatGPT, then Copilot, Cursor, Anthropic, Claude Code, and now Codex. every week there is some new framework, tool, devtool, library, architecture, or agent workflow. everyone is running in the same direction, trying to move faster, ship faster, think faster.

and honestly, that part is exciting.

but watching code appear on a screen is not the interesting part anymore.

the interesting part is what happens before that.

what do you notice? what do you care about? what problem keeps pulling at you? what tiny idea in your head refuses to leave?

because now, if you can think clearly enough, you can bring a lot of things to life in a few minutes. that is still magic. as a creative person, this is the part i love. i can look at something, feel that it should exist differently, and almost immediately start shaping it.

but that also means we have too many choices.

we can build almost anything now, but at the cost of affordance. the harder question is no longer, “can i build this?” the harder question is, “should this exist like this?”

that is where the real work is.

the joy is not in vibe coding. the joy is in solving the problem. it is that small happiness when something finally clicks. when the thing in your head becomes real. when it saves someone time, saves them energy, helps them make money, or simply makes their day easier.

that happiness is hard to explain.

but it is the whole point.

agents are useful here because most work is not sacred. a lot of it is just repetition, coordination, cleanup, checking, moving things from one place to another. agents should take more of that away from us.

not so we can stare at more generated code.

so we can spend more energy on judgment, taste, care, and the few decisions that actually matter.

vibe coding was a good door.

but the room after it is more interesting.