The increasing pressure of climate change and global water shortage requires a major rethink of agricultural irrigation practices. Ensuring optimal agricultural yield while preserving water resources has become a crucial issue for the sustainability of agricultural systems. The management of this resource in Morocco is of paramount importance. This also has repercussions on agriculture, given that a large portion of water is dedicated to this activity. In this perspective, our study proposes an innovative architecture of a four-layer intelligent irrigation system, exploiting the technologies of the Internet of Things and artificial intelligence. The system is based on four interconnected layers to optimize agricultural management. The first layer, dedicated to detection, uses IoT sensors to collect real-time data on soil moisture, temperature and precipitation. This information is then transmitted via a network layer that ensures reliable connectivity through communication protocols. The third layer ensures data storage and processing using artificial intelligence algorithms, making it possible to generate accurate forecasts on key indicators such as plant water stress and evapotranspiration. Finally, the application layer transforms these analyses into actionable recommendations, accessible to farmers via an intuitive web GIS platform. This IoT-AI architecture brings significant improvements to irrigation practices in drylands by optimizing the use of water resources through smart and real-time management. It enables more accurate decision-making by integrating data from IoT sensors and artificial intelligence models, thereby reducing water waste while maximizing agricultural yield. In addition, it promotes increased resilience to climate and economic challenges, thus providing a sustainable and efficient solution for modern agriculture.