The widespread adoption of SaaS platforms has revolutionized industries, offering seamless collaboration, scalable workflows, and efficient resource management. However, managing the dynamic and encrypted network traffic generated by these platforms presents significant challenges for traditional security tools. This paper proposes an innovative framework that combines machine learning with network traffic analysis to address these issues. By leveraging AnyProxy to intercept and decrypt HTTPS traffic, the framework extracts essential data such as URLs, headers, and content types. Middleware components process this data into structured datasets for real-time classification using advanced machine learning models. With 98.6% accuracy in SaaS service classification and 99.8% accuracy in activity classification, this method provides robust differentiation of SaaS applications and their activities. Through feature extraction from HTTPS traffic and machine learning, this framework enhances visibility into SaaS interactions, thereby improving security and compliance monitoring in cloud ecosystems.