The problem with building a company around such a solution rather is this: Is there a market? Sure, having a machine automatically answer customer requests is a nice parlour trick but what's the benefit for the customer?
Just saving money on call centre agents? Or does the software actually solve a real problem in a novel way that can't be tackled using existing solutions?
The benefit for the company looks obvious to me: save money on call centre / community management. Deliver a service to millions of mobile users without requiring them to download your app (just adding a phone number to the agenda).
The benefit for the end-user: having a conversational interface with a system is probably the lowest barrier entry to that system for a human. Everybody knows how to chat using whatsapp, but not everybody knows (nor is willing to learn) how to use your "user-friendly" GUI.
I'm using Python/Django and Yowsup (a python lib that reverse engineered whatsapp's protocol) with PostgreSQL/Redis.
I wonder what you guys think about building a company around this idea. I have a prototype and a couple of customers, so the market is there, but I don't know if whatsapp could sue me or something.