Call routing using artificial intelligence

A tangible, non-transitory, computer-readable medium, comprising computer-readable instructions that, when executed by one or more processors of a computer, cause the computer to:

• request a machine-learning prediction of likely motivation for an automated call between a caller and an automated telephony system, by providing, to a machine-learning analysis system, data from the automated call;
• request a machine-learning prediction of a stress level for the automated call based on speech characteristics of the data from the automated call and an indication of external activities occurring within a threshold period of time with respect to the automated call;
• and
• dynamically route the automated call based upon the machine-learning prediction of the likely motivation, the stress level, or both, by:
• when the stress level is above a high value threshold, automatically forwarding the caller to a human representative, independent of a user input indicating feedback associated with the likely motivation when the stress level is above the high value threshold;
• when the stress level is above a medium value threshold, placing the caller in a medium priority position of a wait queue to reach the human representative;
• and
• otherwise, placing the caller in a low priority position of the wait queue to reach the human representative.

A computer-implemented method, comprising:

• requesting a machine-learning prediction of a likely motivation of a caller by providing, to a machine-learning analysis system, data from an automated call;
• requesting a machine-learning prediction of a stress level based on speech characteristics from the data from the automated call and an indication of external activities occurring within a threshold period of time with respect to the automated call;
• and
• in response to the requests, dynamically routing, by adjusting automated call prompts, the automated call based upon the machine-learning prediction of the stress level and the likely motivation, by:
• when the stress level is above a high stress threshold, automatically forward the caller to a human representative associated with the likely motivation, independent of a user input indicating feedback associated with the likely motivation;
• when the stress level is above a medium stress threshold, place the caller in a medium priority position of a wait queue to reach the human representative;
• and
• otherwise, when the stress level is below the medium stress threshold, placing the caller in a low priority position of the wait queue to reach the human representative.