I’ve noticed a shift in how I code. I rarely write code anymore - claude-code does most of the coding for me. I sculpt the outputs into something more durable.

Once the code is in a working state, based on my original prompt/specs, I then proceed to sculpting - remove excess lines, delete bloat, make sure the pieces fit and can be reused. I study the diffs. I think about what we are trying to implement, and if this iteration brings us closer to it.

Where things go, how pieces fit, reusable patterns - this is more question of subjective taste and big-picture thinking.

I don’t think the coding models we have can replace humans - yet. Proof of that is I still need to write prompts for anything to get done.

Today’s coding agents are not good at being creative. At being bold and coming up with novel solutions. They don’t do well at making critical judgement calls, pointing out egregious flaws.

Where they shine is regurgitating beautiful coding abstractions that have stood the test of time - code I often wouldn’t be able to write myself. This isn’t new - I would never have written my own sorting algorithm to sort a list in the past. I would instead rely on abstractions left for me by those with more experience.

Models build on top of what’s there, and will find all sorts of ways to add code to make it work.

They almost never remove dead code. They rarely tend to disagree with me. They rarely tend to say no.

Without a sculptor, this leads to a bloated statue with no defining features, too heavy to stand, no story to tell. For now I chip away at the code.