The coupled atmosphere-ocean model AWI-CM has been augmented for ensemble data assimilation using the parallel data assimilation framework (PDAF). AWI-CM consists of the atmosphere model ECHAM6 and the unstructured grid finite element ocean model FESOM. PDAF provides the environment for ensemble forecasts and the ensemble filters for the assimilation. The work aims at strongly-coupled data assimilation, hence using cross-covariances between the atmosphere and ocean in the analysis step of the data assimilation process. As a first step oceanic observations are assimilated into the coupled model system in a setup of weakly coupled data assimilation and the effect one the coupled model state is assessed. We discuss the setup of the system, which is generic and hence also applicable for other coupled, but also uncoupled models. Further, challenges of the assimilation into the coupled system and initial results from strongly-coupled assimilation are discussed.