Real-time frequent pattern mining for business intelligence systems are currently in the focal area of research. In a number of areas of doing business, especially in the arena of supply chain management systems, real-time frequent pattern mining is in need. The need is being felt more due to the possibility of real-time knowledge discovery along with the gradual acceptance of technologies like RFID and grid computing and the huge amount of possibilities they promise for real-time decision making like supply-chain optimization. In this paper, we describe a domain-independent heuristic, h1-max and a heuristic search algorithm, BDFS(b)-h1-max for real-time frequent pattern mining, even using limited computer memory. Empirical evaluations show that the techniques being presented can make a fair estimation of the set of the probable frequent patterns and completes the search much faster than the existing algorithms.