Abstract
<p>Chatbots or virtual agents are getting popular in many domains, for example, in academics, business, or healthcare sectors. They are tailored for specific scenarios, and current studies show that they help in reducing the workload of human experts. This study aims to present an architecture and design for a chatbot named as ‘Leo’. Leo is designed using psychological theories to provide support to people who are seeking emotional support. An experiment was conducted to evaluate the usability of Leo, in which 220 participants provided their feedback after having a conversation with Leo. The selected participants had anxiety and mood related issues reported in the past. Multiple Linear Regression was applied to predict the effects of chatbot towards Multiple Linear Regression was applied to predict the effects of chatbot towards supporting the users by regulating their mentalstates. According to the results obtained from the Multiple Linear Regression, using lefta positive influence on the mood of the users, and they recommended Leo.</p>