TL;DR - The major difference is that while Shortcat uses the accessibility API, KayZeer uses image processing. My idea is that you can use it whether your apps support that API, or even if you are working in a VM.

Shortcat does have the advantage that you can fuzzy search. KayZeer at the moment does no OCR or any of the sort.

On the other hand KayZeer (sorry for the bad name, Kay was my labs name, and "seer") had stuff like "." repeats the last command, and it has a LRU of the last "point" you clicked on. You can also manually add points which is useful for instance for each monitor I add 3 points across the screen which normally tie to a file/project view and 2 editor tabs.

But I do accept (and welcome!) suggestions.

I tried going the ML route but boy.. I just couldn't get it to be accurate and/or fast. Maybe I'll retry someday.

KayZeer also seems much faster (I am not surprised because I tried using the accessibility API and gave up because while is sounds efficient in practice it felt very heavy).

Wow you had me spooked. I'd never seen shortcat before and it just so happens it has the same default key bindings as KayZeer, it is also a "cat" while mine is a "dog". So the moment I trigerred the key binding I got spooked because it did exactly what KayZeer.