Multiagent Systems consist of multiple computing elements called agents, which in order to achieve a given objective, can act on their own, react to the inputs, pro-act and cooperate. Data Mining deals with large data. Large data clustering is a data mining activity wherein efficient clustering algorithms select a subset of original dataset as representative patterns. In the current work we propose a multi-agent based clustering scheme that combines multiple agents, each capable of generating a set of prototypes using an independent prototype selection algorithm. Each prototype set is used to predict the labels of unseen data. The results of these agents are combined by another agent resulting in a high classification accuracy. Such a scheme is of high practical utility in dealing with large datasets.