Although adaptive thermal comfort models are incorporated in international standards, field-based evidence linking locally calibrated comfort thresholds with measured HVAC energy performance remains limited in hot semi-arid climates. This study presents a one-year field investigation conducted in an administrative building in Errachidia, Morocco, representative of a hot semi-arid (BWh) climate characterized by high solar radiation and large diurnal temperature variations. The study is based on continuous monitoring data, including hourly environmental measurements and 15-min interval energy consumption records, occupant comfort surveys, and cooling electricity measurements. The analysis identifies a locally calibrated neutral comfort range of 22-26 degrees C, with an adaptive upper comfort limit approaching 28 degrees C and tolerance up to 30 degrees C under behavioral adaptation. The results also reveal asymmetric thermal responses and hysteresis effects, indicating that occupants' thermal perception depends on recent thermal history rather than steady-state temperature conditions alone. To evaluate the operational implications of adaptive comfort, a data-driven regression model was developed using measured building operation data to estimate cooling energy demand under adaptive temperature control scenarios. Scenario-based evaluation suggests that applying adaptive temperature setpoints could reduce cooling energy demand by approximately 21-25% during peak summer months and about 21% annually, while peak cooling load and HVAC operating hours may decrease by approximately 47% and 7-8%, respectively. By experimentally linking adaptive comfort calibration with measured energy performance, this study provides new field-based evidence supporting climate-responsive HVAC operation strategies for buildings located in hot semi-arid environments.