For capture sharpening, I tend to prefer DxO PhotoLab which has profiles for each supported camera and lens combo. It seems to be able to apply different sharpening point spread functions between centre and borders of the image, which is especially useful for zooms.
Yes, that's a benefit. It would take a lot of work to do it ourselves, with sharpening layers. The only downside, besides the cost, is that it uses a generic model for the lens. Individual lenses vary somewhat.
Another approach is taken by e.g.
TopazLabs Sharpen AI, which uses machine learning models that have been trained to recognize blurred features in all sorts of subjects in an image, and replace them with unblurred versions of it. It's also very effective in removing motion blur or camera shake, even in complex non-linear movement. It's not perfect in all images (and good quality input images still produce the best result), but it usually improves the image quality significantly.
It's interesting that, based on user feedback, most images seem to benefit from that (non-symmetrical) motion blur correction. In fact, that became the default AI model but there are user options that allow to change models on a per-image basis. They are now also improving the masking capabilities that allow deblurring to be confined to the most relevant parts of the image.
