This paper explores a two-stage approach for specifying and optimizing feeder bus services for high-capacity rail stations. First, an agent-based simulation is run to identify the service that a new service would generate. Second, an Integer Linear Program is used to construct a bus line in order to satisfy the identified demand. The methodology has been designed to integrate closely with the agent-based transport simulation framework MATSim. The presented use case is based on the Grand Paris Express project in the Paris region, with a specific focus on the Paris Saclay area. The results show that the method is suitable for designing feeder lines for future Metro stations.