[trinix912]

> Most of the "LLMs are slowing me down/are trash/etc" discussions I've had at work usually come from people who are not great developers to begin with - they end up tangled into a net of barely vetted code that was generated for them.

This might be your anecdotal experience but in mine, reviewing large diffs of (unvetted agent-written) code is usually not much faster than writing it yourself (especially when you have some mileage in the codebase), nor does it offset the mental burden of thinking how things interconnect and what the side effects might be.

What IMO moves the needle towards slower is that you have to steer the robot (often back and forth to keep it from undoing its own previous changes). You can say it's bad prompting but there's no guarantee that a certain prompt will yield the desired results.

[forty]

That's my feeling as well, I work on a fairly large and old code base I know pretty well, and generally Claude doesn't build things really faster than I would and then I spend more time reviewing. I end up using it for the most boring tasks (like dumb refactoring/code reorganization) where review is dumb as well and try to have it work when I'm not myself coding (like during meetings etc), this way I never lose time.

[noo_u]

It's definitely anecdotal - and I agree about steering the robot. I find that analysis is harder than creation usually.

[XenophileJKO]

I think that is the skill that separates agentic power users from others.

You have to be really good at skimming code quickly and looking at abstractions/logic.

Through my career, I almost never ask another team a question about their services, etc. I always built the habit of just looking at their code first. 9 times out of 10, I could answer my question or find a workaround for the bug in their code.

I think this built the skill of holding a lot of structure in my head and quickly accumulating it.

This is the exact thing you need when letting an agent run wild and then making sure everything looks ok.