The website [1] is worth reading too.

I like the idea of an automated prover and an SMT solver running concurrently, one looking for a proof and the other for a counter example.

[1] http://cosette.cs.washington.edu/