Creating Digital Twins requires a multi-stage process: The simulation pipeline begins with creating a model of the phenomenon of interest. As a next step, discretization in space and time is necessary. Finally, we implement the solution algorithm in software that lets us compute an approximation of the behavior of the physical system. The coupling library preCICE supports the implementation of simulation software for multiphysics problems and Digital Twins. It follows a partitioned approach that combines highly optimized and specialized tools for single-physics simulation to solve a multiphysics problem. However, the high flexibility of the partitioned approach also comes with potentially high diversity in the complete simulation pipeline. In this setting, developing robust software that produces accurate results is challenging. This talk overviews how preCICE allows to combine heterogeneous software components that use different modeling- and discretization techniques: I show how preCICE allows the coupling of PDE-based simulation models to system simulation models (i.e., Simulink or Modelica), how we can combine different discretization techniques and resolutions in time and space, and how we can conveniently combine software written in different programming languages through the preCICE API.