I would suggest parallel efforts that feed each other. Researching verticals can help the overall design of .WRK.
Having clients/verticals in mind can help focus technical development by finding out what people really want before building it. Designing for specific contrasting verticals can help flesh out true generalized functionality. It can also lead to partnerships with actual clients/users that can help .WRK adoption down the road.
We agree on what needs to be done. I'm just suggesting an additional technical step (adding pre-packaged configuration functionality) and essentially a formalized process of tracking user scenarios (usually a design step) that allows us to reuse the information to build the pre-packaged configurations.
I want a generalized .WRK. One that can easily be used by several verticals without having to be refactored. We already have a lot going for us with .LRN and the other things you mentioned.
