Comment by atleastoptimal

This is very cool, but it's gimmicky. All of the rich text could simply be a modifier before or after the word (such as an adjective or phrase). Given most LLM work is plain text, this benefit isn't as neatly transferable as prompt engineering.