I will introduce the (3+1)D dilute Glasma approximation. This novel approach is a semi-analytic framework for the computation of rapidity-dependent early-time observables, such as the energy-momentum tensor, in relativistic heavy-ion collisions. The Glasma is a model for the earliest stage of a heavy-ion collision. It emerges from the Color Glass Condensate effective field theory, which describes the gluonic interaction between the static sources localized in the colliding partners with classical Yang-Mills equations. We then linearize the Yang-Mills equations assuming weak sources and obtain remarkably simple, analytic solutions for the Glasma field strength tensor. Throughout the calculation, we explicitly break boost-invariance by respecting the longitudinal extent and structure of the colliding nuclei. This allows us to describe the longitudinal dynamics of the Glasma, which is very hard to achieve with traditional lattice-based methods. I will discuss the ramifications of the longitudinally extended collision region that results from breaking boost-invariance. I will present our results for Pb+Pb collisions at LHC and Au+Au collisions at RHIC energies using a generalized, three-dimensional McLerran-Venugopalan nuclear model with parametrized longitudinal correlations. I will analyze how adding sub-nucleonic “hot spots” affects the final energy distribution. Finally, I will demonstrate how we can calculate gluon numbers in the dilute Glasma framework and contrast results for different nuclear models that are matched to transverse momentum distributions.