Data practitioners often need to quantify uncertainty when generating forecasts. However, most popular approaches to forecasting, like Gradient Boosting Machines (GBM), only provide point predictions. To address this limitation, the AI for Retail Lab in Amsterdam recently released Probabilistic Gradient Boosting Machines (PGBM), a PyTorch/Numba-based framework designed to help users apply probabilistic forecasting techniques in large-scale industrial settings (and to solve other large-scale tabular probabilistic regression problems). PGBM, which also offers high-quality point estimates, generates these probabilistic predictions with a single ensemble of decision trees, thereby avoiding excessive computational costs.