In recent decades, computational analytical approaches have revolutionised research in urban planning and design, offering new insights that traditional approaches omitted. Combining social science approaches with cutting-edge digital technologies, these developments help to simulate and forecast human behaviour in urban public places. Among them, Space Syntax constitutes a substantial framework; Visibility Graph Analysis (VGA) serves as an important tool for investigating socio-behavioural dynamics in the urban planning process. However, enquiries concerning the empirical validity of these simulations remain a critical issue. This work intends to expand the dialogue on validating simulation-based results using empirical research. It particularly scrutinises how agent-based simulations, guided by VGA’s parameters, match observable human movement and socio-behavioural activities. The case study is “Bastione di Saint Remy”, a historically and socially important urban square in Cagliari, Italy, noted for its energetic pedestrian bustle and attractive setting. This study aims to validate simulation tools by comparing agent-based simulation results with real-world mobility patterns, thus addressing gaps in their applicability and reliability. The methodology employs Space Syntax principles, utilising Depthmap software to conduct simulations, while human mobility patterns are analysed through systematic empirical observations. Although visual integration and visual connectivity show partial alignment with observed pedestrian movement, visual control and the visual clustering coefficient indicate a stronger correlation with human mobility and static behaviours. Furthermore, the study reveals underused areas within the square, therefore emphasising the necessity of further research on activating leftover spaces.