Innovative shared mobility services involving electric autonomous shuttles have increasingly been implemented in recent years. Due to technical and legal restrictions, these services are usually implemented on fixed circuits (e.g. city centres, hospital campuses, airports) and operate according to fixed schedules. This research studies a variant in which the shuttles’ stopping patterns and schedules are determined in a flexible manner, based on real-time user requests. The aim is to determine the shuttles’ stops, schedules and recharging scheme, such that the total user journey time and the total number of performed laps is minimised. Experiments based on a real-life system in Renmark demonstrate that such a service allows providers to reduce operational costs and improve service quality.