Abstract
A models@runtime environment keeps a model in synchrony with a running system, this way a reasoning engine adapts the system by modifying this model. Existing models@runtime environments typically fail to let the user control what concepts form the model nor how the model is synchronised with the running system. This is yet mandatory in uncertain environments that are open, dynamic and heterogeneous. In this position paper we evolve the classical models@runtime architectural pattern to address this issue, together with some initial implementation results.