Based on a review of supply chain management and intelligent agent theory, this paper establishes a multi-agent technology as a tool for solving strategic selection of supply chain management. It also constructs a supply chain framework based on MAS. Secondly, emphasis has been put on the combination and work content of data-agents, dynamic forecast-agents and decision support-agents. Moreover the operation mechanisms of supply chain framework and vendor status of framework are also discussed. Furthermore, this paper makes use of data-mining technology and genetic algorithms to optimize system learning ability as well as strengthen the self-determination and sociality of the system to find the appropriate vendors.