Solving the science data downlink problem for Mars Express (MEX) has been a challenge. An artificial intelligence (AI) based tool, MEXAR2, has been developed and integrated in the mission planning process that allows the user to quickly and effortlessly generate and analyse a dump plan for any planning period. MEXAR2 is the first AI-based tool in an operational environment on-ground at ESOC and it is the successful implementation of a prototyping study performed before the launch of MEX. The paper explores the challenging requirements levied on the MEX science data downlink, the different solutions for generating a dump plan before implementing MEXAR2. It further describes in detail the interfaces and functions of the software and provides an evaluation of the operational benefits from using the application compared to previous solutions.