A unique QR code is generated for the From: field. It primarily has the unique postal address/business you are wanting to send the goods from. It also has any other details, time, date etc so that it is unique to that package at that time, but is still defined as from a single address.
The To: field is also a QR code which has the address details and similar unique information embedded into it.
Each letter then only has two QR codes, no written text, or even stamps on the face. This is because the postal service could provide the back end which generates the QR codes. You pay for one code to an address that isn’t yours and it deducts it from your account. You print/stick it onto your letter and it has all the details on it required.
As I understand it machines already sort our mail with optical scanners for delivery addresses, and where it’s illegible it gets hand sorted. By removing the manual hand written aspect, there would be no manual processing required. This could even let you track (with limited precision) the progress of your letters. Letters are scanned at the distribution centre and logged as being processed and then again when in the location distribution centre for the delivery area. The post office back end would know who had sent what, and where, and be able to tell you where it had been processed last.
Because most mail is already optically processed, it could even be integrated into international mail processing facilities without the need for too much extra hardware.
At some point a written address would need to be added at the processing plant so that it isn’t misplaced when it gets delivered by a human. This could be done instead of printing a postmark over a stamp, though.
Because the QR codes would be unique, they wouldn’t be able to be used more than once (the backend would retain the history of sent items) but i’m not sure how viable that would be.
Anywho, that’s all my brain could figure out while lying in bed last night trying to think of sheep and fluffy pillows.