The grading of fresh milk affects the quality classification in the dairy industry. This study aims to analyze and design a smart grading system using machine learning models to classify the grade of fresh milk. Business process analysis helped understand the capturing steps as the main elements, such as the smart grading system. The result of the requirement analysis showed how smart the grading system involved stakeholders. The machine learning model can help the Internet of Things system classify goods or services. Artificial Neural Network and K-means were designed to classify and group indicators of fresh milk quality. The variables used in this study consisted of pH, temperature, odour, turbidity, colour, fat, and taste values. The data were taken from the upstream dairy industry SAE Pujon. The classification result of fresh milk grades using ANN consisted of three low, medium, and high grades. The accuracy value of the classification obtained is 98.74%. The attributes used for grouping were temperature and colour. The best clusterization that used K-Means is the third cluster. Based on the data analysis, the smart grading system made users save time knowing the grade of fresh milk easier.