@kristoph yeah i am having trouble w the idea of scale necessary for continuous stream processing of all data emissions into a single monolithic AI... and if not that, then there's at least some other thing going on to preprocess / index / retrieve relevant context