We present a new multi-agent approach to managing how requests are sent between users of smartphone-based applications for reaching bilateral agreements. Each agent is modelled as having a selfish behaviour based on his preferences and an altruist behaviour with respect to the links between the agent and his neighbours. The objective is to maximise the likelihood of an acceptable match while minimising the burden on the users due to unnecessary messaging. We provide a dynamic algorithm using this architecture and we present an empirical evaluation with various mathematical models of user behaviour and altruism. The evaluation shows that our approach can reduce the risks of rejections and the number of requests while increasing the likelihood of acceptable matches.