In commercial spoken dialogue systems, call flows are traditionally built by call flow designers with a predefined business logic. This talk presents a
method for automatically deriving a call flow, minimizing the average number of user turns given a business logic and a frequency distribution of call reasons. As an example, the method was applied to a call routing application whose manually built call flow is processing about 4 million calls per month and whose call reason distribution served to measure the impact of the automatic call flow generation.

When the number of possible caller intents reaches a certain point in applications such as call steering, voice search, and FAQ, few enterprises can afford the required months of utterance analysis, grammar development, and tuning. This session presents an innovative solution to the problem of handling complex caller intents. A dialogue manager automatically generates the call flow from a model of possible intents. Examples discussed  include a large natural language application deployed in the travel industry.