#outdated #idea ![[Pasted image 20210612122009.png]] Basically, people have an audio call, recorded and "analyzed" by AI, and that try to understand the knowledge exchanged and then can display it in various forms, sort of automatic note-taker? # Technical details - Need speech to text (easy) - Extract knowledge (more difficult but doable, can do named entity recognition, classification, and stuff with semantic search, similarity search, i.e. transform text into numbers representing their semantic thus allowing to see how similar "the cat is black" from "the dog is black") - cons: - can be costly in computing? - Displaying knowledge into a proper UX, actually can be hard? not a UX expert! (LOUIS) I guess a graph could be nice, like Obsidian do maybe # Businesses doing it - https://otter.ai/ -> its just transcription, no intelligence behind it, no analysis, no knowledge, just information