Project management requires coordination of tasks, deadlines, dependencies, resources, and risks. Manual status reporting and inconsistent prioritisation can make emerging delivery problems difficult to identify. This paper proposes and demonstrates a framework connecting task prioritisation, progress monitoring, and task-delay risk identification. The framework combines transparent rule-based scoring, monitoring flags, a machine-learning classifier, and human review. A prototype was evaluated using 1,000 generated task records, with a Decision Tree compared against a majority-class baseline on a held-out test set. The Decision Tree achieved 61.0% accuracy, 27.4% precision, 57.5% recall, and an F1-score of 37.1% for the delay class. The baseline achieved 80.0% accuracy but detected no delayed tasks, demonstrating why accuracy alone can be misleading in imbalanced classification. These results describe the prototype experiment and do not establish effectiveness on real project records. The paper concludes that an integrated framework can provide a reproducible basis for further evaluation, with real-world validation required before operational use.

Keywords: artificial intelligence, project management automation, task prioritisation, progress monitoring, risk identification, delay prediction, Decision Tree, business analytics