In-silico simulations such as free energy perturbation calculations (FEP) are an indispensable tool in drug discovery for predicting protein-ligand binding affinity. More recently there has been a focus on approaches that combine FEP calculations with machine learning (ML) methods to broaden the scope of applications and reduce the computational expense. Herein, we describe a workflow that combines a variety of in silico library enumeration approaches with an active learning (AL) accelerated FEP protocol capable of predicting the affinity of 10’s of thousands of compounds in less than 1 week. User success stories including experimental validation are presented and opportunities for integration with parallel synthesis and ML ADME prediction are discussed. This methodology represents a transformative approach for comprehensively evaluating chemical space and optimizing small molecule binding potency.