South Korea's food delivery platforms are using artificial intelligence and operational data to improve service accuracy as competition intensifies. Baedal Minjok, also known as Baemin, will introduce a system next month that automatically predicts delivery times and informs customers. Restaurant owners previously had to estimate and enter cooking and delivery times themselves. Baemin's model uses 58 data types, including restaurant-level delivery conditions, area congestion and rider availability. The company said its forecasts are accurate more than 50% of the time, exceeding the accuracy of owner-entered estimates by more than 1.7 times. Rival Coupang Eats has developed store-coordinate optimization technology that analyzes actual pickup points, delivery partners' routes and nearby road information to guide riders to restaurant entrances more precisely. The company said inquiries about store locations and orders taking more than an hour to deliver have declined since the technology was introduced. The industry considers reliable arrival estimates and accurate pickup navigation important to customer satisfaction and repeat use, particularly as free-delivery promotions attract users but delays can drive them away. Baemin's monthly active users rose 5.8% year over year to 24.4 million last month from 23.06 million, while Coupang Eats increased 23.0% to 14.44 million from 11.74 million. A retail-industry official said reducing uncertainty in the delivery process is important for satisfaction and repeat usage, and that service quality will become a core competitive factor as rivalry intensifies.