Even prior to Bill Gibson's blog on the devarchitect, my own thoughts on personas have been that they are very useful abstractions that help us focus us on the target user and hence prioritize features accordingly. It is essential that we create those abstractions correctly. Consequently, Bill's hypothesis of the devarchiect can have serious implications for the future direction of our product.
However, until our product has been in the market for a few years, I am not sure how good a predictor presonas are for who our eventual users are going to be. It is entirely plausable that our tools might even make architectural tasks more accessible to developers. Given Visual Studio's strong developer ("programmer") following, I wouldn't be surprised if the actual users using our tools end up being (senior) enterprise developers. In fact, I personally think that some of the tools that we have built might be very useful for developers - an example is the code-generation and synchronization functionality.