This study aims to explore the practical path of deep learning in the smart classroom environment, so as to fill the gap in the current research on flexible deep learning in the smart classroom. In terms of methods, we adopted the consistent design concept, combined with modeling methods, educational design research and educational experimental research, constructed a deep learning architecture model for intelligent classroom, and designed a deep learning design framework and deep learning scaffold. Through expert consultation, literature search, questionnaire survey, trial, and gauge evaluation. The results show that we have successfully defined the deep learning architecture for the intelligent classroom, constructed the corresponding design framework and bracket, and obtained the practical results through educational experiments. The experimental results show that the proposed deep learning scaffold can effectively promote students’ deep participation, the adoption of learning strategies and the development and transfer of high-order knowledge in the intelligent classroom environment. In conclusion, this study provides effective solutions and theoretical support for the practice of deep learning in the intelligent classroom, fills in the gap in the current research, and has important theoretical and practical significance.