The "genetic" part of their pitch comes from the way you build things. You're given a screen of 10 different choices and you choose 2. Then the "genes" of those choices are mixed and mutated and you are presented with 8 more choices, the 2 you chose last time and 8 new "children". Through a series of choices you eventually narrow down your choices until you decide you're done and you assemble your robot. In theory you'd think this would work, but in practice I had a hard time intelligently guiding and massaging a particular trait from the choices. I think that's because Dr. Fluff uses by-the-book genetic algorythms which are known to have a problem with variable length traits being extracted from the genetic data stream. All I wanted was a top fin, but I kept getting a wild array of robots that look nothing like their parents. But my kids didn't mind that so much when they played with it and settled on whatever they were shown pretty quickly.
netfabb and careful alignment on the build plate was necessary to have a chance at success, and at that I was only able to make them look good from the front. From the back where all the supports attached is an different story (the picture I took of the prints from the back didn't do justice to the horrors presented).
Over all I'd say Dr. Fluff is promising and I'm going to keep my eye on their next app, but Modio is way better for my purposes and makes better toys (though I haven't managed to make them print just yet). I just wish that Modio ran on Android or Windows.