Causal dynamical triangulations (CDT) is a sum-over-histories approach to quantum gravity which leverages the techniques developed in lattice quantum field theory. In this talk, I discuss the thick sandwich problem in CDT: Given initial and final spacelike hypersurfaces, each with a fixed geometry, what is the transition amplitude for one transitioning into the other? And what geometries dominate the associated path integral? I discuss preliminary studies performed in this direction. I also highlight open problems and interesting directions for future research.