A method, article of manufacture, and apparatus for automated processing of an auditory communication is disclosed. In an embodiment, this comprises receiving the auditory communication with call transport information, triggering a first policy based on the data, and selecting, based on the first policy, at least a first lexicon comprising a plurality of keywords. A first portion of the call is compared to the first lexicon to identify keywords matched by the first portion of the call, and a second policy is invoked based on the matched keywords. The action(s) specified by the second policy is taken. Actions may include termination of the call, joining a third party, archiving the call, changing a characteristic of the call, or further processing using another lexicon.