rocess conditions, and finished tofu attributes. When supported by reliable historical data, machine-learning tools may assist with recipe adjustment, anomaly detection, predictive maintenance, and early identification of deviations in texture, yield, or coagulation behavior.
The practical value of AI depends on instrumentation, data quality, cybersecurity, and operator oversight. Sensors must capture relevant variables such as temperature, flow, dosing, pressure, and equipment condition. Human validation remains essential because soy-milk composition, coagulant chemistry, and product specific