Holistic understanding of well operations can play a key role in optimization and maintenance of assets. The traditional process for understanding well operations mainly involves fitting a curve through all of the historical production data and extending the curve to forecast production without modeling the stochastic nature of production history, considering impact of any well interventions, or feeding any a priori information into curve-fitting workflows. This leads to unreliable production and reserves estimates which, in turn, impact the strategy and planning process for asset management. A novel workflow was developed learns the production characteristics of a well through a statistical framework using principles of signal processing and Bayesian inferences. Using this workflow, high-fidelity empirical production performance forecasts can be obtained for all the wells in the asset in an automated fashion. This novel workflow aims to significantly simplify interpretation of well operations, reduce the turnaround time for analyzing and modeling well performance, and improve the quality of reserves estimates.