A multi-agent systems approach to call-centre management

Abstract

The (human) intuition when modelling a hierarchical business process is to reflect the incipient control pattern in the agents that comprise the physical organisation. One rationale behind this approach is that stakeholders can relate their perception of the business process to its software implementation. We describe a hierarchical approach to building a model of a call centre, in which the key elements are a call-router and a variable number of call-handlers and note that the communication overheads and complexity of the agents required to implement the call routing process rapidly increase with the sophistication and scale of the call-centre model. We contrast this with a non-hierarchical approach building on ideas from peer-to-peer networks and demonstrate through simulation the similarity of the key performance indicators derived from each approach. We conclude that the inherent flexibility and resilience of the decentralised approach suggests significant potential advantages compared with current approaches to call-centre management, and looking beyond this scenario, for dynamic task allocation in general. Furthermore, the multi-agent solution is capable of functioning both as an application and as a simulation that then allows experimentation with alternative business models and potentially enabling more effective day-to-day allocation and utilisation of resources within actual call centres.