Hi, there are no LLMs involved, it is all local and an embedding (vector representation) of the data is created and then that is used for search later, nothing is sent to cloud from your files and there are no local LLMs running as well, only the encoders (I use the Perception Encoder from Meta released a few weeks ago).