The paper addresses the problems of efficient representation, maintenance and exploration of social knowledge enabling task decomposition, organization of negotiations, responsibility delegation and other ways of agents’ social reasoning. We focus on multi-agent systems for integration of already existing software components. It is supposed that all the social knowledge is kept sepa-rated from both the problem solving knowledge and agents’ specific internal in-telligence and that it is organized and administered in the acquaintance models located in the agents’ wrappers. A specific tri-base acquaintance model (3bA) is formalized and discussed throughout the paper. This model helps to optimize the communication traffic, to implement meta-reasoning processes and supports the machine learning activities. Several practical applications of the 3bA ac-quaintance model in different fields are presented and the acquired experience is discussed.