I think this is just creating the layout files, all the work of wiring it up still has to be done by the programmer. This is really cool and probably a glimpse of the future, but honestly it would probably be easier to train the designer on how to use the drag and drop UI builders that already come with these platforms.
That assumes the designers want to learn them. Sure, some of them are happy to, but this negates the need for them to do anything other than concentrate on their area of concern.
Why train someone to do grunt work when the machine can do it?
Translating picture designs into html files is not really that much fun.
Now the next step (probably more difficult) is you need a way to let the programmers "hook" into the generated output in order to further tweak it or customize it, but _without_ having to modify the generated files.
It looks like a very good first step. I wonder how structured/semantic the generated HTML would be for engeniers to "hook' the generated code with a backend
