Transformers are the most vital parts of any electrical supply system. The reliability of a supply system is largely dependent on the reliability of the transformers. Thus, timely fault diagnosis in a transformer can contribute to the robustness of the electrical supply system. Available techniques for diagnosis of abnormalities in a transformer are part specific and do not detect complete range of all possible faults. Also, different techniques may not be concurrent while detecting the same fault. To overcome this issue, three popular techniques which cover a wide range of faults are integrated using multi-Agent systems (MAS). These techniques are modeled as agents using Java Agent Development Environment (JADE) platform to make them capable of communicating among themselves to ensure effective decision making. The goal is to design an approach enabling detection of problems in transformers which can lead to accurate prediction of faults without continuous human monitoring. A tool with a graphical user interface (GUI) is also made using java programming.