Based on Artificial Intelligence technology, personalized learning is poised to be an important direction for university course reform in the future. Addressing the issues of insufficient practice opportunities, delayed feedback, and challenges in providing personalized feedback in the digital skills practice teaching of the “New Media Marketing” course. This study aims to design a system module that provides an intelligent practice environment and automated personalized feedback. Drawing upon theories of personalized learning, Intelligent Tutoring Systems (ITS), and automated feedback, and integrating key technologies such as Natural Language Processing (NLP), Machine Learning (ML), and Generative Artificial Intelligence (LLM), this study proposes a conceptual model and preliminary architecture for an intelligent module targeting complex digital skills practice in New Media Marketing. It describes the core logic and feedback generation algorithm framework for evaluating complex, open-ended practice outcomes within the intelligent feedback engine, highlighting how AI technology can be utilized to parse, evaluate unstructured creative text and strategies, and generate multi-dimensional, explainable automated feedback. The module design and algorithm framework proposed in this study offer novel and feasible ideas and technical solutions for addressing the feedback challenges in practice teaching.