Watson out of the walled garden
Every Watson deployment I've covered thus far has been, essentially, a pilot-like deployment in a training phase with lots of care and feeding from IBM and carefully selected partner customers. The partners usually also have a vested stake in success. Once Watson Analytics goes live within one month, we'll finally see the vast public -- or at least the vast IBM customer base -- banging on this technology and trying to get smart results out of it.
Although "Watson" is on the outside wrapper, most of the data-management and data-analysis technology here really isn't part of the Watson Cognitive Computing technology. I take the natural-language-processing interface to be the main contribution from Watson. The main giveaway here is that training is not required, although some of that training is built into the "story" templates for specific types of analysis.