You can add all kinds of fancy stuff, you can improve the design and usability of the UI's but here's a big important problem waiting to be solved: audio quality and graceful degrading of the quality when needed.
At the end of the day we want to be heard in the first place. Nothing is more irritating than someone's (or your own) poor connection that makes people ask each other to repeat, watch participants freeze, wait, waste time trying to reconnect etc. etc.
Can it be solved? Degrade the quality, compress more but deliver the human speech as nicely as possible, on time. I feel not all options have been tried in this area yet.