Abstract
Fermentation endpoint prediction is important in brewing because incorrect timing can affect attenuation, flavour stability, tank scheduling, and batch consistency. This study examines AI-assisted endpoint prediction using pH, density, and temperature profiles collected during beer fermentation. Controlled ale fermentations were monitored under fixed wort composition, yeast strain, pitching rate, oxygenation, and fermentation temperature settings. Density decline, pH reduction, temperature variation, sugar consumption, ethanol formation, attenuation, yeast growth, and fermentation duration were evaluated as model inputs. The results show that combined pH, density, and temperature profiles improved endpoint prediction compared with fixed-time fermentation schedules. Density provided the strongest indication of sugar conversion, while pH and temperature improved early detection of abnormal fermentation behaviour. The AI-assisted model reduced uncertainty in endpoint estimation and supported more consistent process decisions. The study demonstrates that profile-based fermentation analytics can improve brewing reliability, tank utilization, and final beer quality.