What would you do if you were a system at criticality confined in a bounded domain? Of course you would forget about details of the interaction, and lattice spacing, flowing to an RG fixed point. Besides attaining this bulk universal behavior you would also try (boundary condition permitting) to forget about the confinement becoming "as uniform as possible". Implementing this requirement in absolute geometric language, the one used by general relativity, we obtain novel predictions for the structure of one- and two-point correlators. These predictions are tested successfully against numerical experiments yielding a precise estimate of a critical exponent of the Ising model in three dimension. New preliminary results for the three dimensional 3d xy model will also be presented.