This work investigates the use of enterprise models for service configuration with a focus on knowledge related to organizational roles. Starting from an industrial case, it presents an approach to extract a role’s information demand from an enterprise model and to use this as initial configuration for agent-based services. In addition to the meta-model included in the enterprise modeling language, the use of a common ontology is proposed which captures both, the perspective of information demand and services. The architecture of the agentbased services is following the idea to characterize all actors by their roles and to represent them by sets of services. This approach facilitates self-organization in the service level. The main contributions of the paper are (1) to show that a role’s information demand is relevant for service configuration, (2) to present a way of extracting information demand from enterprise models, and (3) to extend an approach for SOA-based enterprise representation with information demand.