We propose an agent-based grocery shopping system, which automates grocery shopping process by comparing grocery information based on user's preference. The agents gather grocery information from several store server agents and compare it with user's preferences of groceries and stores. The agents could adapt to user's up-to-date preference by learning from his/her evaluation of previous shopping results. There are three organization agents in our system: a user agent, an information management agent, and a store server agent. The organization agent is responsible for organizing the primitive agents and coordinating message delivery. By implementing a prototype of the agent-based grocery shopping system, we have confirmed that it is instrumental in purchasing the best possible groceries of user's preference from several grocery stores. This system eliminates or reduces the burden of grocery shopping, and helps user save enormous time required shopping around grocery stores.