[whpalmer4]

I think it would be completely unworkable before OmniFocus 1.1 ships, with its support for synchronization between computers. On the other hand, one could probably use OmniFocus 1.0 in a solo trial run to see if the basic tool provides the sort of functionality that you need. If it does, then you know that OF 1.1 is worth waiting for to investigate further; if it doesn't, then it's time to look elsewhere. If it looks like OF 1.0 is a promising lead, then sign up for the 1.1 sneaky peeks and expand your trial run to include another member of the firm to test the multi-user dynamics. I wouldn't trust my professional workflow to the 1.1 sneaky peeks to the exclusion of all else; track your efforts in parallel until it is clear that you understand the system well AND that it is sufficiently reliable. Maybe start out with just a fraction of your caseload that is well defined so that you can "walk before you run" and not have to spend too much effort remembering "oh yeah, the Palmer case, that's one we're tracking in the new system as well."

One wrinkle of using a GTD-like system in a multi-user environment is that you'd probably want some way of indicating that someone has started work on a given task, so that you don't have multiple people looking at the available actions and selecting the same one on which to work! Maybe an entry in the notes field with a time/date stamp (cmd-| will get you one) would do the trick, depending on how quickly the synchronization software propagates the change. Of course, maybe accidental duplication of work isn't an issue and everyone stays out of each other's way, and you just need the system to be able to see what is going on. Might still want to do the notes trick just to see that a given task has been claimed and started, rather than marking something as complete when in fact you're just starting work on it. Or maybe you change the context of a task to "In Progress : Associate LH" when LH starts work on it; presumably the real context is no longer relevant at that point.

It's certainly an interesting experiment, and I for one would like to hear more about how it works out.