AI coding agents changed how I write code, but not how I think.

Over the last year I've been relying heavily on coding agents.

They've dramatically reduced the amount of code I type manually. Boilerplate, refactoring, tests, documentation—they're all much faster now.

But one thing hasn't changed.

The hardest part of software engineering is still making good decisions.

Choosing the right architecture.

Understanding trade-offs.

Knowing when not to automate something.

I've noticed that the better the engineer, the better they seem to use AI—not because they write better prompts, but because they ask better questions.

I'm curious whether others have experienced the same thing.

Has AI changed the way you think about software engineering, or just the speed at which you write code?