The aerospace industry continuously improves surface geometries through high-fidelity (HiFi) simulations that couple several disciplines, notably computational fluid dynamics (CFD) and structure mechanics. Achieving a target accuracy demands very fine meshes, especially for high-Reynolds-number aerodynamic configurations. A powerful mean to accelerate these simulations is refining the mesh adaptively only where it is needed. We propose a fully CAD-driven mesh adaptation workflow that is tailored to linear hybrid structured-unstructured meshes commonly used in industry. The CAD model serves as a reference to guarantee that the adapted mesh remains faithful to the originally designed geometry. The workflow proceeds as follows: 1. CFD solution: A next-generation, highly parallel CFD solver [1] computes the flow field. 2. Local refinement: Driven by the CFD solution, the adaptation library FSAdaptationNG refines cells locally. Inside the boundary layer mesh, refinement is performed on entire cell stacks, maintaining the anisotropic layering in a form essential for robust CFD and the subsequent mesh deformation. 3. CAD-conforming boundary correction: The corrected boundary-node coordinates are computed by FSOCCT and supplied to FSMeshDeformation, which deforms the volume mesh so that it conforms to the CAD geometry. All steps are orchestrated within the FlowSimulator [2] environment, integrating various high-performance computing (HPC) components with focus on avoiding bottlenecks. The framework exploits MPI parallelism and works in memory, achieving highly scalable performance on modern supercomputers. We present the application of the process chain to three-dimensional test cases that mimic realistic aircraft configurations. Compared to global-refinement techniques, the adaptive approach delivers the same accuracy with significantly reduced cost. Moreover, we discuss the HPC efficiency of our approach, highlighting the scalability of the involved packages. Our results verify that CAD-based, stack-preserving mesh adaptation is an industry-ready tool for accelerating HiFi simulations while maintaining geometric fidelity and computational efficiency.