One way to reduce the empiricism associated with open die forging (ODF) design is through finite element (FE) simulation. While this approach provides a more reliable way to estimate the final geometries, process variables, and mechanical properties of the part, its multi-step simulation might require significant computational and time efforts to deliver the numerical outcomes. This challenge finds a suitable solution through machine learning (ML) surrogates, as patterns can be identified within the FE outputs, the forging variables are correlated with some mechanical properties distribution, and numerical simulations can provide a large dataset to build high fidelity artificial intelligence models. This work proposes modeling regression ML surrogates for the ODF finite element simulations using methods such as random forest, artificial neural networks, and principal component analysis. The focus is on predicting local equivalent strain and geometry of the final workpiece, with aluminium alloy AA7010, friction, and ingot’s dimensions as input parameters. By implementing mesh manipulation techniques, different algorithms, and hyperparameter optimization this work demonstrates the feasibility of achieving a final predictor that delivers similar results (±0.1 as standard deviation) to the FE equivalent strain outcomes and is around 2000 times faster than the traditional numerical approach. Additionally, the deployment of the final surrogate in a graphical user interface allows for intuitive implementation of the model in the Aerospace unit of Otto Fuchs KG.