General circulation models based on flexible, unstructured discretisation strategies offer an opportunity to explore a range of multi-scale flow phenomena and planetary climate processes. The formulation, implementation and  configuration of such frameworks is rather non-trivial though - providing  numerous avenues for new research. In this talk, I present recent work on the  development of the NASA-GISS ocean model (GO) and the Model for Prediction Across Scales (MPAS-O); exploring the use of unstructured and/or flexible grids  in both the horizontal and vertical axes of the models. I focus on the  development of new algorithms for the construction of 'optimal' Delaunay  triangulations and Voronoi tessellations, hybrid vertical coordinate systems, Arbitrary Lagrangian-Eulerian techniques, and the formulation of 'mimetic'  unstructured discretisation schemes. I introduce two new, open-source software packages designed to support unstructured GCM development and explore opportunities to interface with the ICON modelling community.