I disagree. My ability to write sufficient code to patch prior issues hasn't led to a loss of velocity. Basically, run a bunch of agents in parallel and they should be able to smooth out any issues.

https://imbue.com/product/mngr/

Started from a thread on "Slop is not necessarily the future" (https://news.ycombinator.com/item?id=47587953). I commented that technical debt from sloppy code shows up too late to fix. someone replied that humans also write sloppy code. Fair enough, but that got me thinking about what's actually different when AI writes the code.

Short version: AI doesn't lower the bar for code quality. It just makes the lack of process hurt faster. Two things get amplified: the gap between what you ask for and what gets built, and not knowing which generated code is going to survive.