Been watching Mycroft for a couple years now, have been very excited at the idea of running an opensource AI at my home. However…digging deeper recently it appears I made a bad assumption: I’d thought one would download and run the Mycroft backend on one’s own server and point one or more client devices at that server.
But that’s wrong isn’t it? One can download the Mycroft client to a range of hardware, but it must always be connected with Mycroft’s centralized service, right?
Assuming that’s correct, is this architecture a short term approach? Might running the backend locally be available some day…or is that simply not practical?