It'll end up being specific to the species of bacteria causing the patient's infection, which means they'll first have to isolate the source of infection. Sometimes this will be an easy task, sometimes not. I suspect over the long term treatments that go to market will combine several (or many) natural or engineered phages into a single delivery protocol and those that don't find their natural targets will die out harmlessly.

This is a super interesting area of research that I think will ultimately complement treatments like biota transplants for chronic diseases and antibiotics for acute disease. As a therapeutic approach (at least in the US) its still young though, so I wouldn't expect to see dramatic changes for years yet.