As demand for air transport continues to rise, sustainable flight operations demand an increasingly dependable departure time to reduce air delays at congested airports and minimise negative environmental impacts. We propose an artificial intelligence (AI)-based recommendation system for assisting pilots and airline operators in selecting appropriate departure timings. The system is designed to anticipate and suggest departure schedules that minimise delays by utilising historical flight data and taking into account the interdependence of many flight departures. The proposed approach uses an agent-based technique to analyse the complicated relationships between flights and generate optimum departure plans. The system uses both individual flight times and cumulative effect of concurrent flight operations to achieve comprehensive optimisation. The proposed AI-driven recommendation system enhances departure planning, resource allocation and departure time recommendation using an agent, achieving a root mean square error (RMSE) of 8.67 min overall, with XGBoost demonstrating the highest accuracy in flight duration prediction (MAE: 5.18 min; RMSE: 6.85 min), outperforming competitive models like LightGBM and CatBoost. This AI-powered recommendation system will improve operational efficiency, reduce fuel consumption and associated emissions, and contribute to enhancing sustainable air transport operations.