The present-day forecasting processes rely heavily on statistical models developed directly from source data, guided by input from subject matter experts. This approach, while functional, is inefficient due to its high operational costs, challenges in maintainability, and dependence on individual expertise. To overcome these limitations, the proposed work presents the development of a unified, global forecasting model leveraging AI and machine learning technologies. This paper focuses on forecasting revenue at the segment level and predicting units based on product type, region, and combination. The research work employs an iterative and incremental approach to develop a model for heavy construction industry sales and revenue data, supplemented with external economic drivers. This work emphasizes data protection and explores optimal forecasting methodologies. The LSTM Model demonstrated the highest accuracy, achieving 96.8% for Segment Revenue forecasting, 95.5% for product type forecasting, and 95.6% for region-based forecasting. Additionally, the combined region and product type model showed optimal performance using LightGBM, with an accuracy of 92.4%.