I think it very much depends on what you're working on, and how much you're responsible. For a web front end, not really. A backend where a single line can make a mess out of your data? Yes. Part of the issue too, is code review brings up things you the prompter didn't think about that another teammate may be thinking about. The value goes beyond making sure the code is "correct", but also that it makes sense in your domain.

The "we don't read/write assembly anymore" argument doesn't fit though. Given an input x we know without a doubt we'll get output y. With an LLM, that isn't the case no matter how many guard rails, specs, markdown files, diagrams you give it.

I just had a long session planning an implementation with Opus 5.5 in a reasonably complex system. It was a frustrating affair where it suggested very complex solutions that, sure would work, but would be very slow and difficult to maintain with the rest of the solution. After a long discussion with many variations of complex designs, we found a very simple solution. Elegant even.

I think the problem in this case was: 1. Unusual constraints, that kept nudging the model to overdesign or misunderstand parts of the solution at every turn. 2. One code-comment it found and thankfully mentioned, that a previous session had made up and left there as a fabricated requirement. The model trusted this "requirement" more than my instructions, which unfortunately made sense, since it was indeed written and part of the code.

I'm working on a system that has a mix of trust boundaries with a few perspectives that are a-typical and it trips up the models working on it. Fable 5.2 didn't do better at all.

I have tried to put clear instructions in an agent architecture file, but I haven't managed to get it fluent yet.

If using llms is like using wordpress instead of writing the website yourself probably never

I think we are approaching the point where it no longer makes sense to have LLMs write code in an intermediate, human readable programming language.

I know this topic is very contentious but it seems very clear to me that this is the long-term direction of writing software. UI code isn't quite there yet, which is why we've got all these vibeslop websites proliferating now that all have identical crappy UIs. But I imagine it will get there too.

>Only exception might be the upper echelons of optimization, like language authors or compiler level work. But even then…

Optimization work is especially suited for LLMs. They're great at hill-climbing problems and have infinite patience to test different ideas.

You still reading code?

I stopped reading the code many months ago, probably 18 months ago. I the "early" days of agentic coding one did feel dumb when telling the agent for the 30th time to fix a bug and it just couldn't. Now I cannot possibly imagine a scenario where an agent wouldn't be able to fix a given issue on 1st or 2nd try.

im not reading it. and im using fable up front earlier then having sonnet/opus implement and verify.