solve continuous embedding by using content hashes and last indexed timestamp. the key to preventing a lot embed requests was to only embed once the user opens the AI search modal or toggle the RAG button on ChatNodes. #rabbitholes