My personal impressions is that with their own programming language boo-straping on C++/Boost they are setting themselves up for lots of interoperability problems later. ADDED: To quantify that: I give 70% that they will drop that language completely or fail altogether.
Seems like they are targeting the old relational database coders, and giving them a chance to incorporate some machine intel, to re-examine their data with different tools.
Looks like they are also leaving different paths to insert inquiries, so that they can backtrack if they have to later.
Pretty interesting model..