This video introduces the Flownex ROM (Reduced Order Model) Builder, a tool that uses machine learning and neural networks to create simplified models that accurately replicate the behavior of complex Flownex simulations. The generated models are exported as Functional Mock-up Units (FMUs), allowing them to be integrated into external software environments without requiring Flownex to be installed. The ROM Builder trains neural networks using sensitivity analysis data generated from varying simulation inputs and outputs across defined operating ranges. Once trained, the reduced-order model can be validated by comparing its predictions against the original Flownex simulation results to ensure accuracy. The video demonstrates this workflow using a simple pipe network example, highlighting how ROMs can significantly reduce simulation time and computational requirements while maintaining reliable predictive capability.