We want to start by trying to predict what to reply.

E.g. if the user says: "Sweet!", "Thanks", "Ty", "Amazing", etc. asume that the bot has to reply with "No problem" or "Anytime".