I've been doing this for almost 15 years now. I definitely create new things, but I didn't start that way. And really, you should use conventional things for most problems that you come across. It takes a pretty unconventional problem to justify an unconventional solution.