Right now processes are queried on-demand one-by-one. If expected usage is many processes, would be better to load them all and cache.