Belief-Desire-intention (BDI) agent based systems have been implemented in many business application systems and found to have some limitations in obverting environmental changes, adaptation and learning in making rational decisions. Our paper presents a new hybrid BDI agent architecture which compares all the available intentions in the intention reconsideration process and is able to observe all the events which are related to the committed intention, before a decision is being made. Limitation in capturing of one event in the intention reconsideration process is overcome with the introduction of our extended BDI execution cycle. Further, the use of Knowledge Acquisition Modules (KAM) in our proposed model improves the learning ability of the generic BDI agent. Execution of plans for a committed intention is based on the reinforcement learning techniques and Adaptive Neuro Fuzzy Inference System (ANFIS) is used in deciding the intention reconsideration of the proposed agent model. This enables the agent to interact with the environment more closely and use intelligence in making rational decisions, whose behavior may be not known at the design stage.