Coarse-grained simulations of turbulence must somehow become turbulent. For real engineering problems, this generally means the simulation needs to capture laminar-turbulent transition. This is challenging, because transition is a highly non-equilibrium process, that is not represented by standard theories of turbulence. For turbulent flow, the coarse-graining approach splits the flow into larger scales, which are resolved, and smaller scales, which are modeled. This approach is not generally applicable to transition, which is typically dominated by a single characteristic scale. For example, a Tollmein-Schlichting wave cannot be decomposed into large- and small-scale components. A coarse-graining strategy must therefore, necessarily, either resolve all the fluctuations in the transition region (i.e. DNS) or model them completely (effectively becoming a Reynolds averaged Navier-Stokes RANS model). The first approach is typically used in LES type models, where the early stages of the transition region are fully resolved and the modeled scales are created only as the turbulent cascade develops. The second approach is more typical of hybrid-type models, where the model is allowed to shift to RANS mode in the transition region. This presentation will review the current state-of-the art for coarse-grain simulation of transition, and discuss potential future directions. The effectiveness of resolving transition in LES methods, and the challenges of smoothly switching on the LES model will be discussed. For RANS models, the various types of transition-sensitized models will be reviewed, as well as their application for hybrid modeling. Finally, the talk will conclude with a discussion of ongoing challenges and next steps in transition modeling for coarse-grained simulations.