When building an Intelligent System you’ll need to decide where the intelligence should live. That is, where you will bring the model, the runtime, and the context together to produce predictions—and then how you will get those predictions back to the intelligent experience.
The runtime could be located in the user’s device, in which case you’ll need to figure out how to update models across your user base.
The runtime could be in a service you run, in which case you’ll have to figure out how to get the context (or features) from the user’s device to the service cheaply enough, ...
