If they can do that, then they have found a magic bullet for one of the biggest problems in software development. I have a suspicion that's the real problem they're trying to solve, so they're "doing things that don't scale" to try to engineer a platform that solves for that problem. That's the only way VCs would throw money at a business model like this.

I wish them luck, but I just think there are too many variables that getting a large enough sample size would bankrupt them before they approach a workable solution.