I guess the language models have an advantage here because they don't need to understand the TLAs and "internal lingo" of McCompany X, they can just fake it goo

I guess the language models have an advantage here because they don't need to understand the TLAs and "internal lingo" of McCompany X, they can just fake it good enough (like everyone else does in orgs). But in highly technically important meetings, y'all might want to proofread in case the wrong TLA gets used.