Additive manufacturing (AM) is becoming increasingly relevant across industries due to its tool-free fabrication and high geometric design freedom. Powder bed fusion of metals using a laser beam (PBF-LB/M) as a subcategory of AM enables the production of complex metal components with excellent mechanical properties. However, the process is characterized by steep thermal gradients and a non-uniform cooling, leading to distortions and residual stresses. In severe cases, crack formation and propagation occur. Finally, inhomogeneous microstructures, leading to varying mechanical properties within a part, can arise. These challenges still prevent the PBF-LB/M process from a wide-ranging industrial application. To address these issues, this paper presents AscentAM, a simulation tool based on the finite element method and artificial intelligence (AI) specifically designed for the predictive modeling and modification of the thermal, the mechanical, and the failure behavior in PBF-LB/M. Built on open-source software, AscentAM employs a sequentially coupled thermo-mechanical approach that captures process-induced thermal and mechanical interactions. The tool fully replicates the PBF-LB/M process chain, including heat treatment, and integrates additional functionalities such as a part pre-deformation algorithm and crack prediction. Furthermore, it extends its capabilities to incorporate thermal predictions of optical tomography data and AI-based homogenization techniques for uniform microstructures and, therefore, consistent mechanical properties. The accuracy of the predictions was validated through an experimental benchmarking of distortion, residual stresses, and crack formation in various application-oriented geometries, demonstrating a high level of predictive reliability. The pre-deformation sub-module successfully minimized form deviations, contributing to an improved first-time-right manufacturing. The AI-based prediction of optical tomography data proved to be highly accurate, with the automatically executed laser power adaptions exhibiting homogenized thermal signatures and grain structures. By enabling these precise and computationally efficient process predictions, AscentAM provides a valuable tool for researchers and engineers working on the simulation-driven enhancement of PBF-LB/M.