Market liberalisation and increasing demands for the allocation of services which span several networks are pushing every network operator to evolve the way of interacting with peer operators. The Inter-domain Demand Allocation (IDA) process is a very complex task for several reasons: there are different actors involved, end-to-end routing must take into account QoS requirements, network resources and information are distributed, etc. In this paper we address the problem of the QoS-based multi-domain routing, and more specifically a multi-agent paradigm for supporting the IDA process is defined. We show how Artificial Intelligence methods for distributed problem solving supply a compact way to formalise the multi-domain routing process, and how this formalism enables an agent middle-ware to route demands across distinct domains.