Open sheosi opened 2 years ago
A way that they can be implemented right now is with an initial answer + notification later. However, this opens the possibility for the user to do something mean while, or to look confused as they think the assistant won't answer their question (if their client stops showing the animation of the petition being processed).
With the new serverless skills architecture, we are sending notifications, si we can just send two notifications, maybe we can signal somehow that something else is coming.
Fortunately, there's kind of a solution with the new architecture. Since server->skill vap requests are actually published in a resource which the skill "observes" and any answers are calls from the skill to the server sending "notifications", we only need to send two of those, maybe with a little different structure.
Alexa has a pretty nice feature called progressive responses, here they can say something like "Hey, wait a minute I'm looking for what you said" it might be intieresting to support this in the protocol