At the moment, the roadplan is what I call "swabbing the deck"..
Fixing any critical bugs or functionality so badly implemented that Discourse can't be used properly.
Supporting the early Discourse adopters, those setting up test and trial forums, and those contributing code, on GitHub and here in a "teach the teachers" manner, so they can help others in turn.
Making sure our own house was in order so that we can begin to support our 3 partners as outlined on http://www.discourse.org/buy/
Ensuring that each partner is not just pleased with Discourse but ecstatically happy with Discourse. So far we've implemented one partner, How to Geek, and another is coming before the end of May. We are laser focused on working toward getting our key 3 partners fully implemented and happy.
Prioritizing support for functioning, active installs of Discourse whether we host them or not. Feature requests based on real live user data is always the best kind.
The only practical advice I can think at the moment of is to avoid attempting to add any major new features, I know we certainly are avoiding any new feature work as we stabilize and test the code base in a public environment! That said, small tweaks, fixes, and refactorings are certainly more than welcome!
We'll have more guidance as we get further in, but for now, we've been happy with the pull requests we've gotten.