Since I'm building a very focused piece of HR software that hiring managers and recruiters can use to manage sourcing of resumes and other recruiting activities, I've been watching the various vendors like LongJump and Coghead, who promise a software silver bullet: delivering software without the expense of developing it. I'm not convinced they are going to deliver on that promise.
In some respects, I'm sure the applications they will help their clients build will be fine -- they'll do one thing (perhaps even doing it well), and cost very little to build, and everybody will be happy. In other ways, though, I think those applications will fall short, simply because the thought, the planning, and the refinement that go into a software application built by designers, developers, and domain experts working together with their respective skill sets just won't be there.
As I've been working on Catch the Best, I've been reminded how an idea for an application can start out very small, but the deeper you dig into it, the more complicated it gets. Of course, this is one reason why trying to preserve simplicity in software is so hard. But on the flip side, your application can't be so simple it isn't useful. This can be a fine line to walk in order to deliver excellent software, and just handing someone some building blocks and saying "go build it yourself" will have mixed results.
In other words, sometimes the just-add-water brownie mix gets the job done. If you want the best brownies, though, it needs a little more effort than that. :)