We've been going at this whole "talk to the machines" thing backwards.
It's time to put the back end first.
Systems in the online world are massive and complicated. There are more of them than anyone can possibly keep track of. And many of them are hard to tell apart. We need them to be able to identify and describe themselves so that we can know where to connect.
Every online system has its own internal logic. They've all been built with particular purposes in mind, based on assumptions about how, where, and why they'll be used. Once we've found them, we need to know what they want us to do and how we should go about doing it.
Online systems speak in machine languages. Once they can tell us who they are and what they do, we then need a way to communicate back and forth with us, the humans. We can make this happen by teaching online systems to talk to our personal assistants: Alexa, Google, Siri, ...
Online systems have power and potential far beyond humans. They don't need us hanging around watching while they work. We need to show them how to manage the work they're doing for us from start to finish on their own, in the background. They can always ask if they need help.