The implementation today is as a task queue which removes the request from the queue once a responder acknowledges, but it could be a pub-sub model, where a number of independent responders can work on the same message in parallel, and only 1 responder need to return a response. In this case, persisting in the queue is useful.
An alternative is to chain the responders where one responder can leave a message in the queue for another responder, and the final responder returns the response.
Polyglot is still experimental though, and the current implementation is a prototype.
It wasn't quite dynamic (it required an engineer to set new values for how many workers you wanted..) but we could do this via a GUI.
For (1) what do you do with the persistence? A web request, in general, is not important after a few seconds.
For (2) how does Polyglot accept multiple responders for a single request, and how would it join the responses?