Add on-device document scanning with ML Kit or VisionKit - edge detection, auto-capture, PDF output - then run OCR with google_mlkit_text_recognition.