With the development of electronic commerce, online shopping is becoming more and more popular. It is of great importance to enhance online shopping experience whether for the merchants or for the customers. Online shopping assistant based on multi BDI agent has been studied in this paper. The system architecture was proposed and the BDI agent class diagram was depicted. To study the communication mechanism among agents, the format of KQML message was designed by using Java. The running results shows that this system can alleviate the burden of human-computer interaction, and can meet the customers' demands for personalized and proactive service.