This article presents a unified, automated workflow for artificial neural network (ANN)-based modeling of microwave components using an automated model generation (AMG) method with electromagnetic sensitivity. The proposed method employs adaptive data sampling and automated model structure adjustment strategies to enhance modeling efficiency. During adaptive sampling, EM sensitivities are combined with interpolation models in lieu of repeated ANN training, enabling efficient identification of smooth and nonlinear regions in the design space while reducing the number of required samples and EM simulations. Parameterized modeling examples for a bandstop filter and a waveguide bandpass filter illustrate the application of the AMG framework.