The primary criterion for assessing heating, ventilation, and air conditioning (HVAC) systems regarding thermal comfort is whether they are capable of satisfying more than 80% of occupants (i.e., acceptable condition). The predicted percentage of dissatisfied model proposes this value with the assumption that a neutral state is desired. However, recent studies cast light on personalized thermal comfort which demonstrates that occupants have diverse thermal preferences and respond differently to variations in temperature (i.e., thermal sensitivity). This study aims to shed light on the importance of taking thermal sensitivity into account in a multi-occupancy space, where the same thermal condition is shared, for thermal condition optimization, which was replicated in our multi-agent based (MAB) model. Each human agent (occupants’ proxy) has its own properties (e.g., thermal preferences) and aims to create at least an acceptable condition for itself by providing feedback to a HVAC agent (HVAC systems’ proxy). The HVAC agent optimizes the thermal condition based on feedback from human agents. Using this model, two operational scenarios have been explored: human agents have (1) the same (i.e., ignoring thermal sensitivity) and (2) personalized thermal sensitivities. The assessments demonstrate that integrating thermal sensitivity results in significantly different setpoint temperatures, increased thermal satisfactions of human agents and the number of satisfied human agents. In other words, thermal sensitivity is an important factor in improving the performance of HVAC systems.