Abstract: Most approaches for solving partial differential equations rely on generating a mesh that captures the geometry of the model. At present, unstructured mesh technology is the dominant choice, making it possible to create three-dimensional meshes with hundreds of millions of elements within minutes. Yet, when performing design optimisation, multiple simulations are required for varying operating conditions and geometric configurations. Producing a high-quality mesh for each case is extremely time-consuming, as it demands substantial human input and specialised expertise. In this presentation, I will discuss our recent research on applying artificial intelligence to predict near-optimal meshes for simulation purposes. The central idea is to exploit the wealth of industrial data already available to improve the selection of an appropriate spacing function, including anisotropic distributions. The method seeks to transfer knowledge from past simulations to guide the meshing process more effectively. I will evaluate this approach in terms of prediction accuracy, computational efficiency, and sustainability, considering the carbon footprint and energy demands associated with parametric CFD studies across a range of flow conditions and angles of attack.