I also work on a small team that leverages LLMs heavily. The verbosity and "polish" of the responses is tiring. Everything starts to feel written in just one, inhuman voice. I too miss the "roughness" or personality of writing and communication pre-LLM.
I use this default context in ChatGPT and it works pretty well for me:
"Tell it like it is. Don't sugar-coat responses. Be concise. Short answers only. No bullet points. One paragraph max. No over-explaining."
And what you will get, of course, is (perhaps convincing enough) performative mimicry of "telling it like it is", kind of like when suburban white people rap.