The lack of digital thermal comfort prediction models that incorporate the dynamic influences of the surrounding environment raises the need for more advanced algorithms, such as those associated with artificial intelligence (AI) and machine learning (ML). The existing outdoor thermal comfort prediction models used in the simulation tools depend on instantaneous grid calculations, despite the influence of antecedent spaces that pedestrians endure, instead of using a dynamic calculation based on the accumulation of factors from their thermal environment. The urge to design pedestrian-friendly environments in warmer cities requires an advanced assessment that considers not only microclimatic conditions, but also the physical, social, physiological, and psychological factors that control pedestrians’ behaviors and decisions. This study introduces a comprehensive thermal comfort assessment model that integrates the most influential factors in warm climatic conditions using actual datasets gathered from fieldwork. Backpropagation artificial neural networks and nonlinear regression models for dynamic thermal assessment of multi-agent simulations were implemented to aid the accuracy of the simulations used for thermal comfort predictions and integration with a walkability scoring system. The accuracy of the ensemble machine learning model reached 79%, and the comparison between different climatic data files revealed logical variations for a 20-min walk for all simulated agents. This study reports the structure of the thermal comfort model, a group of validation cases, and future development prospects.