Sandy soils in Egypt's newly reclaimed lands face multiple challenges due to their low water-holding capacity, nutrient leaching, and high evapotranspiration, all of which threaten sustainable crop production. This study evaluated an artificial intelligence-driven decision support system (AI-DSS) for managing irrigation and fertilization in wheat (Triticum aestivum L.), maize (Zea mays L.), and sugar beet (Beta vulgaris L.) over three consecutive seasons (2022-2025). The AI-DSS integrated real-time soil moisture, nutrient, and weather data using Random Forest and LSTM models to optimize input scheduling. Compared to conventional farmer practices (CFP), AI-DSS led to yield increases of up to 13.1 % and improvements in water use efficiency (WUE) by up to 15.5 %, particularly in sugar beet during 2024. Partial factor productivity (PFP) also increased significantly, especially in maize. Post-harvest soil analysis indicated higher residual levels of nitrogen (+13.6-19.3 %), phosphorus (+22.7-25.0 %), and organic matter (+17.9-22.0 %), along with a 13-19 % reduction in soil salinity. Economic assessments showed an 8.5-15.0 % increase in the benefit-cost ratio (BCR). Additionally, nitrate leaching was substantially reduced under AI-DSS, mitigating environmental risks. These results underscore the potential of AI-driven management to enhance productivity, input-use efficiency, and soil sustainability in coarse-textured soils of arid regions.