Agriculture is the backbone of the Indian economy that employs over 50% of the population and also significantly contributes to the GDP. However, the sustainability of the sector is highly challenged by factors like erratic weather patterns, variability in climate, and most importantly assessment and mitigation of risk. Similar problems exist in traditional insurance models, they fail to assess risk and provide insurance and safety to farmers as much as it is needed. This paper proposes an AI-based model that results in a Dynamic Insurance System to meet the shortcomings of traditional models. This model uses advanced machine learning techniques to enhance risk prediction and insurance efficiency. The model collects from various sources including satellite imagery, weather data, soil health metrics, and historical yield data that is processed through FCN, Vision Transformers, and Temporal Fusion Transformers. Using these technologies crop health prediction, weather impact analysis, and soil suitability assessment are carried out more precisely. The proposed model calculated the risk factor and then that risk factor is included in all the insurance calculation formulae which makes them dynamic and more precise. The comparative analysis shows that Dynamic Crop Risk- AI Model (DCR-A)I, the proposed model) achieves higher accuracy than other models. This research is data data-driven approach that offers a robust solution to risk mitigation in agriculture, enhancing farmer security and promoting sustainable agricultural practices in India.