A Proactive Management Assistant Chatbot for Software Engineering Teams: Prototype and Preliminary Evaluation

Abstract

This work presents a prototype and preliminary evaluation experiments of a management conversational assistant for software engineering teams. Such teams commonly rely on text chat for communications using specialized collaboration platforms such as Slack or Microsoft Teams. A conversational assistant integrated into these platforms can help in several tasks related to project and team management, thus benefiting both managers and software developers. The proposed design of the conversational assistant is described in terms of its architectural design and interaction design, and the important design decisions are explained. Preliminary evaluation shows that the proposed chatbot can be used effectively to interact with users reactively and proactively over public and private channels. However, for better results, a specific intent classifier should be used for each combination of user roles and channel types. In addition, interacting to messages with unspecified target over public channels is affected by the relatively limited intent classifier accuracy and still requires improvement.