This work presents a multiscale modelling approach for the laser powder bed fusion (PBF-LB/M) process, taking into account different material models and various physical phenomena across multiple scales. At the smallest scale, a thermomechanically fully coupled model explicitly accounts for phase transformations, enabling accurate prediction of the microstructure evolution. The integration of transformation-induced strain effects naturally captures layer thickness variations during phase transitions. A simulation strategy for complete parts is proposed, characterized by its computational efficiency and the utilization of micromechanically motivated inherent strains, cf. [1]. This strategy facilitates the prediction of physically well-motivated residual stresses and deformation. In contrast to empirical averaging methods, the proposed thermomechanically coupled framework directly incorporates phase transition between the powder, molten, and solid phases. The material model's flexibility enables its extension to advanced solid-state phase transformations for multiphase alloys, such as Ti6Al4V, using physics-based evolution equations governed by dissipation functions, as discussed in [2]. This approach is both practical and reliable, as minimal experimental data is necessary for the parameter identification of the evolution equations. It enables the precise reproduction of Continuous Cooling Transformation (CCT) diagrams, thereby enhancing the predictions of microstructure-dependent mechanical properties. Overall, the multiscale methodology effectively addresses the challenges posed by different modelling scales (without full scale separation), while maintaining adequate computational efficiency for part simulations. By capturing the key multiphysical interactions of the three different scales, it advances the understanding of the thermomechanically coupled material responses and development of different solid phases in PBF-LB/M. This capability enables the prediction of microstructure evolution, residual stresses, and distortion. In summary, the proposed framework has the potential to function as a robust and extensible tool for process optimization and the design of metal-based additive manufacturing. [1] doi: 10.1016/j.addma.2022.103277 [2] doi: 10.1007/s00466-024-02479-z