This commentary proposes a novel strategy to achieve patient remission through conceptualized “nudge drugs”, or state-shifting drugs, which gradually transition the tumor microenvironment (TME) to one unfavorable to cancer progression. Unlike the typical monotherapy approach to cancer treatment, modulating the TME would occur via a sequential drug therapy model, in which the first nudge drug is a requisite for the second one. Each nudge drug would bring the TME to a mediatory transition state on its path from the cancer state to the remission state. The sequence of drugs would be determined through coordinated gene programs (GEPs), which allows for the description of the archetypal state of a tumor by linking its component cells’ gene expression profiles. This allows for the prediction of how tumor-host interaction networks may shift when a drug is introduced by using deep generative models to create computational atlases predicting the outcomes of perturbation. Additionally, The classification of tumor immune archetypes is of particular importance to this strategy as they allow for the prediction of archetype transitions by the model. Thus, this enables the model to plan the path to a remission state, detailing which nudge drugs and their correlated transition states are suitable. Although the concept is still in its early stages, this innovative approach aims to improve patient outcomes by modulating the TME into a state unsuitable for cancer progression and favorable to remission.