Artificial Inteligence
Intelligent processing of mail has been around for decades. Its roots can be traced back to the US Post Office Department deploying optical character recognition (OCR) letter sorters in the mid-1960s. By the 1980s, the US Postal Service (USPS) was pushing the envelope with the Multiline Optical Character Reader (MLOCR) sorting machine. That was followed in the 1990s with software that could read handwriting.
Vendors began marketing similar solutions to corporate mail centers in the 1990s. The outside of the envelope could be read by a camera, then sorted automatically to the correct department. Bins were set up in “mail run order”, creating further efficiencies.
Integrating scanners with openers and extractors, a new level of automation was introduced. Initial gains in payment processing expanded to inbound and internal mail. Operators are able to open mail, scan contents, classify the document type, and route it digitally to the right person or department. All from a single workstation.
The next step for inbound mail processing is to go from moving images to processing the contents of the correspondence. With intelligent document processing, the content is extracted and processed by an automated workflow. That will be the focus of next week’s newsletter.
Inbound package processing has followed a similar path. The first generation of tracking software matched the address keyed from the shipping label to an internal database. A second label with the appropriate mailstop and barcode was created. Digital signatures were captured on handheld devices.
The current tracking systems send automated notices to recipients, integrate with locker systems, and predict delivery times. AI powered systems recognize variants of names, and store that information for faster processing in the future. Cameras with computer vision can go “beyond the barcode”, recognizing and identifying packages by size, shape and condition.
The top three challenges faced by mail centers when implementing AI powered solutions:
- Prioritization. The corporate mail center is too often in the physical and organizational basements. IT resources and funding budgets are dedicated to other projects.
- Integration. An intelligent mail center may need to talk with an Enterprise Resource Planning (ERP) system, Human Resources platform, facilities management software, and accounts payable (AP) system.
- Training Data. Any AI system needs to be trained on your mail, not generic mail. That means unique processing paths, document types, and limited versions of addresses. Training the system will require up front work by the mail center employees.
The journey to integrating AI systems in inbound mail and packages may be longer than anticipated, with bumps in the road and unplanned detours. By starting today, managers will begin the transformation to the mail center of tomorrow.
See you next week!
Amazing Astronomical Fact: If you have clear skies this Saturday night, February 28th – look up! Shortly after sunset, six planets will appear in the western sky – Mercury, Venus, Saturn, Uranus, Jupiter, and Neptune. Venus and Jupiter will be the brightest, appearing like bright stars low in the western horizon.
