Bob does not write code anymore, he orchestrates agents. Says it makes him faster. That the level of slop is the same as with humans. He swears by it.
Bill only uses agents to write tests and do code review. He says even a Fable 5.1 will create 1,000 line slop instead of his 10 line fix. He swears by it.
Both Bob & Bill are both credible developers that have lots of experience at building real things, including at Google.
Do agents still produce slop? Does the 10x engineer really exist? What is your position?
What's your view?
The skill of reviewing requires experience and competency in the technology. It will require critically thinking around why the agent chose a technology over another and what the "definition of fixed" not workarounds means by the experienced human.
Whether if it is a big name company or a lesser named company is irrelevant.
> Do agents still produce slop?
Yes.
> Does the 10x engineer really exist?
Yes. They do not care about what company they join. They might even build their own company rather than join anyone else and they do not care about titles. (Which is all made up).