The problem of demand-responsive passenger transportation is a subject covered from different perspectives (such as planning, user interfaces, optimization problem, etc). This work is focused on the simulation of events occurring in real time. That is, as requests are generated by vehicle users and the planning moves forward in time events may occur that affect them so that there should be a re-planning procedure. Among the diverse considered events can be considered: user delays and cancellations of requests, failure of vehicles in operation, among others. Under these premises is given a software architecture based on agents that is capable of supporting such events and hence perform the replanning. This is a description of the architecture, design and implementation by using intelligent agents under the PASSI methodology.