Making and answering audio calls - either by phone or an app - would really be the killer app for me. Being able to say “Genie, call Alex” or “Genie, call mum on Element/Skype/Nextcloud Talk” would be amazing.
Obviously, this would need to be built on a couple of foundational skills, namely:
- Access to contacts
- Access to an app capable of making audio calls.
For contacts, I would suggest a contacts skill that connected to carddav, as an open standard for contacts, used by large, self hosted projects like NextCloud and many others. Knowing the users contacts could also open up other options like the early demos of Almond where one server could access a contacts almond server (with permission, of course!)
For calling, you could either have the server pair with a phone by bluetooth, or access a webapp by api (eg Matrix with Element, or Skype for Web)