A real-time audiovisual conversational model that listens, reasons and speaks simultaneously, reading tone, facial expression, gesture and hesitation from live video and generating an expressive avatar response.