In some specific work contexts, such as writing pull request descriptions, not sounding like AI is something I've given up on trying to optimize. It's simply not worth the effort for me being non-native and writing detailed PR descriptions being so arduous, and the agent already has full context anyway. Obviously any fluff or inaccuracies are aggressively weeded out but I don't care anymore about the AI voice.

> any fluff or inaccuracies are aggressively weeded out

this work is paramount. Without clear evidence of human filtering, a long, well formatted message/PR/doc is likely to reduce my estimate of the value/veracity/relevance of its content.