Designing a virtual receptionist Using contextual information and natural interfaces

Abstract

In this thesis, we describe our work in designing and building a virtual receptionist system, called “Amee”, for Interaktionsbyrån’s studio in Gothenburg. Amee’s responsibilities includes welcoming people to the studio and providing services to visitors and employees alike. A significant part of the project concerns how to make the receptionist context-aware and have it interact with the user in a natural human way. To achieve this we built a receptionist system based on a network of microservices that collaborate to create a unique user experience using face recognition and text-to-speech technology. In our work we identify and discuss challenges in building context-aware applications and how publicly available information can be used to enhance the user experience. Going further, we examine how such a system might be received by users through a series of user tests and interviews. Our findings show; that users enjoy being received by Amee, that they have few concerns about privacy in this context, that further study needs to be done in how to manage the quickly increasing complexity of a conversation model. The project contributes to the current body of research as a cursory investigation into the area of designing automated personal service agents for a non-personal context.