This is neat. I wonder… Is there a more comprehensive analysis of how well the Apple Vision framework compares to other multi-modal AI? Would there be any benefit to pre-processing images via Auge before handing them to Claude, GPT?