Industrial laser processes are still predominantly optimized through empirical iteration, limiting scalability and development speed. The ongoing shift towards digital process engineering requires predictive simulation approaches that avoid process-specific calibration and phenomenological model tuning. We present a physics-based framework for laser material processing based on a fully compressible multiphase Mass-of-Fluid formulation. In contrast to conventional approaches, phase transitions and momentum exchange are resolved within a consistent thermodynamic framework, avoiding recoil pressure models or similar fitting parameters. This enables predictive simulations without recalibration, requiring only physically consistent material parameters. The predictive capability is demonstrated for laser beam welding and additive manufacturing. Building on this backbone, the framework is extended with additional physical models, including two-temperature formulations and material-specific absorption, to address short and ultrashort pulse processes. This enables simulation of drilling and cutting across metals, semiconductors, and dielectrics. Validation against experiments highlights the potential as a pathway towards digital process twins applicable in industrial settings.