You may as well go ahead and implement the whisper code now as much of it will still be useful; the message processing and whatever queue setup you end up going with so you don’t smash up against the rate limits, for example. At most, it’s likely that the only change we’ll have to make is how we connect (which is one of the easier parts) and how we encapsulate and the messages which may just end up being some JSON marshall and unmarshall calls.
Either way, go for it! It definitely enhances the user experience by reducing chat volume.