Once that is solidified, the next phase built upon this tech foundation will be a general purpose bidirectional RPC "sublayer" that operates on any medium (low or high bandwidth, reliable or not, requiring encryption or not) and can be trivially built upon for more specific applications.
Once that is done, I'll use it to move up the chain to actual protocols we need going forward.
There are a bunch of other technologies locked up in my brain as well, but I figure I have about 50-100 man-years backlog already...