The estimated number of outpatients with skin diseases in China is ~200 million per year, while the dermatologists are insufficient and the doctor-patient ratio remains low, which causes fewer patients receive effective diagnosis. Compared with others, the diagnosis of skin diseases, which is less reliant on laboratory tests, imaging and pathology, needs the assistance of large hardware devices. By contrast, dermatologic diagnosis requires a combination of visual inspection and interrogation frequently which is exactly what Artificial Intelligence (AI) specialises in — Computer Vision (CV), Natural Language Processing (NLP) and Speech Recognition (SR). This allows a simple image capturing tool embedded with an AI model to perform dermatological diagnosis at the primary level. Hence, based on the dataset, which from Asian, with more than 200,000 images and 220,000 medical records, we explored an AI skin diseases diagnosis model---DIET-AI to diagnose 31 skin diseases, covering the majority of common skin diseases. Ranging from 1st September to 1st December 2021, we prospectively collected case information from 15 hospitals in 7 provinces in China, using mobile devices to collect images and medical records of 6043 cases. Then, we compared the performance of the DIET-AI with 6 doctors of different seniority in the prospective clinical dataset, concluding the average performance of the DIET-AI in 31 diseases is no less than that of all different seniority doctors. By comparing the area under curve (AUC), sensitivity and specificity, we demonstrate that DIET-AI model is effective under the clinical scenario. It is further validated under more complex clinical scenarios, providing references for exploring the feasibility and performance evaluation of DIET-AI in clinical use afterwards