This study aims to implement a web based automatic complaint classification system using the Random Forest algorithm at the Department of Population and Civil Registration of Pesisir Selatan Regency. The research applied the Research and Development (R&D) method, including needs analysis, system design, prototyping, testing, evaluation, and refinement. Complaint data in text form were processed through preprocessing stages (case folding, tokenization, stopword removal, and stemming), followed by TF IDF feature extraction before classification using Random Forest. The model was evaluated using accuracy, precision, recall, and F1 score metrics. The results indicate that the system was successfully developed and is capable of classifying complaints automatically and in real time with good accuracy. The integration of Laravel and Flask API supports efficient classification, while the verification feature ensures category accuracy. The system improves efficiency, accelerates complaint grouping, and enhances public service quality.