Despite being key components of modern-day organizational work flows, coordinating meetings currently requires a con- siderable amount of manual effort, time, and is error-prone. Current solutions typically use either semi-automatic schedul- ing that requires a structure of user input or partially manual scheduling that relies on some degree of human involvement. As such, we propose SmartMeet - an AI system designed for fully automatic meeting coordination based on natural lan- guage commands from its users. SmartMeet leverages several advanced technologies for speech and natural language processing to enable full auton- omy of the system. Among those technologies include au- tomatic speech recognition, natural language understanding via transformers, large language model reasoning, and back- end automation systems. In essence, with SmartMeet users can schedule their meetings through either spoken or written speech and have the system extract all necessary details, find a suitable time period, make corresponding calendar entries, generate meeting URLs, and send automated emails without any human assistance. Based on prior work, the current version of SmartMeet was further developed to improve usability, enhance conflict reso- lution mechanisms, and increase overall integration. Contrary to conventional approaches to scheduling, SmartMeet works using a speech-first method and requires no additional man- ual coordination. Experimentally, it is shown that SmartMeet considerably decreases the scheduling time while minimizing booking errors and improving productivity of the organization in question.
Maddhesiya et al. (Mon,) studied this question.