The present work is related to Web intelligence and more precisely to medical information foraging. We present here a novel approach based on agents technology for information foraging. An architecture is proposed, in which we distinguish two important phases. The first one is a learning process for localizing the most relevant pages that might interest the user. This is performed on a fixed instance of the Web. The second takes into account the openness and the dynamicity of the Web. It consists on an incremental learning starting from the result of the first phase and reshaping the outcomes taking into account the changes that undergoes the Web. The whole system offers a tool to help the user undertaking information foraging. We implemented the system using a group of cooperative reactive agents and more precisely a colony of artificial bees. In order to validate our proposal, experiments were conducted on MedlinePlus, a benchmark dedicated for research in the domain of Health. The results are promising either for those related to Web regularities and for the response time, which is very short and hence complies the real time constraint.