We consider a mix transportation problem, which allows to combine a multi-modal public and a ride-sharing transports, in a dynamic environment. The main idea of our approach consists in labelling interesting nodes of a geographical map with information about either riders or drivers, in so-called buckets. Based on the information contained in these buckets, we compute admissible ride-sharing possibilities. To restrict the needed amount of memory, among the different stops along a public transportation path, we only consider the transshipment nodes, where travellers have to make a change between two modes. Each of those stops are potential pick-up or drop-off stops for ride-sharing. We consider a drivers’ maximal waiting time, as well as the maximal driving detour time depending on the actual drive. Each new drive activates a search for new ride-sharing of existing riders. Each new ride activates another process which searches for potential drivers. Among all admissible ride-sharing possibilities, only those which best improve the earliest arrival time are selected. We provide numerical results using real road network of the Lorraine region (FR) and real data provided by a local company. Our numerical experiment shows a running time of a few seconds, suitable for a new real-time transportation application.