Sorry if this sounds completely dumb but I can't hold it: Can anyone explain what's with this: "you are running ML algorithms in the cloud, and we enable you t

Sorry if this sounds completely dumb but I can't hold it: Can anyone explain what's with this: "you are running ML algorithms in the cloud, and we enable you to run them locally!" Can't we already run them locally??? Why not run them locally in the first place if speed and low latency is required?