want... what exactly would this look like? camera/ eye tracking probably for context + multimodal to see what you're looking at on screen?