In this study, we present CompArt, a novel compartmental modeling framework for crystallization and precipitation processes. CompArt integrates an AI-driven compartmentalization strategy, which automatically determines the optimal number of compartments based on an internal clustering metric (the Davies–Bouldin index), thereby enhancing both accuracy and robustness. The framework also incorporates compartment activation times, estimated using a non-reactive tracer, to account for hydrodynamic delays and spatial variations in reaction kinetics, particularly in downstream compartments. The approach is applied to Mg(OH)₂ precipitation under different operating conditions, considering both purely molecular processes and scenarios including secondary phenomena such as agglomeration. CompArt accurately reproduces the temporal evolution of the zeroth-order moment of the particle size distribution in all compartments, closely matching detailed CFD–PBM simulations while overcoming limitations of standard compartmental models, which tend to overestimate reaction rates in downstream regions due to neglected flow delays. The framework successfully predicts delayed nucleation onset, local growth, and agglomeration dynamics, providing a physically grounded representation of the system. Moreover, CompArt achieves a substantial reduction in computational cost, completing simulations in a couple of hours on a single processor compared to several days on 64 processors required by Computational Fluid Dynamics - Population Balance Model, even when including pre-processing for compartmentalization and flow extraction. These results highlight the potential of CompArt as an efficient and accurate tool for process development, scale-up studies, and real-time applications in both academic and industrial contexts.