Monitoring soil moisture and nutrient levels is crucial for increasing agricultural productivity. Traditional soil analysis methods are laborious and inadequate for real-time decision-making, leading to wasteful resource allocation, which can result in suboptimal crop yields and increased costs for farmers. This research introduces an AI-integrated IoT (AI-IoT) framework for real-time soil monitoring and nutrient forecasting in precision agriculture. The suggested system combines soil moisture and nitrogen (N), phosphorus (P), and potassium (K) sensors with a microcontroller, powered by an off-grid solar unit, facilitating deployment in remote agricultural settings. Sensor data are transferred to a cloud platform via the message queuing telemetry transport (MQTT) protocol, where machine learning algorithms evaluate the data to provide watering and fertilizer recommendations. The system underwent validation at two field sites in Tamil Nadu, India, spanning a duration of 30 days. Experimental findings indicate a prediction accuracy of 92% with minimal error rates, surpassing traditional models. The suggested methodology facilitates data-informed decision-making, minimizes resource depletion, and improves sustainable agriculture practices.