Enterprises are challenging for decision support procedures incorporated in their daily work, for example, recommendation of prices during ordering of commodities in supply process. The practical experience has proved that economical effectiveness may be obtained in enterprise daily supply activities by balancing several factors as priority of customer, reliability of supplier, saleability of product etc. For this purpose the supply manager must be able to evaluate instantly considerable amounts of data. In this work adaptive software agents are analysed intended for using in information systems of such kind. Recommendation system is comprised of six agents for customer, order, supplier, stock, recommendation and data base. Algorithms for computing behaviour of agents are presented. Information system based on agents should make recommendations with regards to prices and discounts in order to support users – supply managers – in their decision making process. Recommendation system is implemented, using Microsoft Visual Studio .NET 2003 and MS SQL Server. Experimental research of system has demonstrated the viability of approach and effectiveness of using recommendation system in enterprise activity.