This work presents a hybrid model for Cabernet Sauvignon (CS) red wine-making that combines mechanistic and data-driven approaches to optimize the fermentation process and improve the quality of red wine. The model incorporates two sub-units representing the interaction between alcoholic fermentation and phenolic extraction, considering factors such as temperature, products addition, draining time, and must composition. To develop and validate the model, a database of 270 industrial CS fermentation from 2017-2021 harvest seasons was collected. The models were calibrated using experimental data, achieving an average R2 of 0.94 for fermentation kinetics model and 45% and 80.9% test accuracy for tannins and anthocyanins predictors, respectively. A multi-objective dynamic optimization problem was formulated and solved to find fermentation operation conditions that optimize simultaneously phenolic quality, process costs and productivity. A similar distribution of the Paretos were obtained for varietal and premium wines. Finally, these tools were packed in a digital platform for practical use in industrial cellars. The models generate the predictions and recipes prescription for each fermentation tank when the pre fermentative juice is analyzed. As a result, it is obtained useful information for wine decision-making like maceration length and wine phenolic composition at least five days in advance.