Perhaps as an alternate keyboard, or some kind of floating button, expose a quick & ready method to begin inputting text into current active cursor from an OCR scan from the camera. There would ideally be a way to optionally have the scan jump to the next column, next line/row, or no jump, after confirming good read (which might be achieved with appending a "tab" or "return" character).
An example scenario: you have been tasked with taking inventory of all the computers in your office, including serial numbers. You create a new spreadsheet on your mobile device. Click in the first row and make a "header" row with labels "office number", "serial number", "item type" and "model number". Next, you walk up to an office, tap on the first data cell (column a, row 2) then press the "OCR Acquire" button (this new feature), then aim camera at the room number sign on the door, click capture button. You are now prompted to confirm the OCR text extraction looks good by tapping checkmark button. That text is entered into cell, and the spreadsheet automatically "tab"s over to next cell. Now repeat tapping OCR input again, but aiming camera at the serial number sticker on the computer, then type in the item type normally, tab, and lastly OCR input again, this time the model number sticker had lots of extraneous text, so you have to exclude that from the acquisition, and then confirm, and then the last cell is filled. Now just tap on the beginning of the next row, head down to the next office and repeat the whole process.
I've seen lots of cool OCR applications, and I know those will put text into clipboard, but the context switch and the extra taps and menu interactions are too time-costly and annoying. Something more streamlined, more like an alternate keyboard/input method that lets me stay "in the application" is best.
This app "Barcode/NFC Scanner Keyboard" by TEC-IT looks like it is almost there. It seems to have more or less the concept, but only scans QR & barcodes, not "plain printed text". Still useful as a demo of how it might work at a high level (special button to take image-based capture).