can someone help if i use screenshare method in vapi sdk.. can ai agent do realtime OCR processing? what is it supposed to be exactly for?