Many vendors are promoting Intelligent Document Processing (IDP) solutions however, there is significant buyer frustration in this market. Many IDP products are failing to live up to vendor claims, with enterprises struggling to automate processing beyond 50-60% of invoices with first-generation IDP.
There is massive interest in extending automated invoice processing to the remaining 40-50% of invoices. This will be accomplished in part by IDP vendors adding reliable handwriting recognition and automating invoice processing for long-tail, low-volume vendors.
Innovative companies, like Merck, have begun exploring higher-level invoice processing automation. In his keynote, Steve Carpenter, Executive Director of GBS Digital at Merck, discussed using machine learning (ML) to identify duplicate and jeopardy invoices (invoices approaching an early discount or payment terms date).
The company’s automation aspirations are high, with goals of: 80% reduction in annual third-party recovery fees, 70% reduction in cost per invoice, 90% perfect PO/perfect invoice/perfect vendor KPIs, and 100% elimination of manual tracking and notification.
Conversations at IA Week gave me the impression that IDP is becoming mainstream technology, entering a rapid expansion phase in many enterprises. Meanwhile, innovators are eagerly seeking to move up the stack and generate greater value by leveraging Unstructured Data Processing (UDP).
Enterprises are beginning to explore automated document processing use cases beyond invoices and purchase orders (PO). We met many conference attendees interested in extracting key information, such as terms or termination dates, from contracts. There was also strong interest in processing supply chain and logistics documents, like bills of lading (BoL).
Innovators, are eager to move beyond IDP to emerging UDP use cases—which involve text, images, videos, and other unstructured data types. A few interesting UDP use cases presented involved using camera footage to automatically identify workers in warehouses or manufacturing facilities that aren’t wearing proper safety equipment (e.g., hard hats, safety glasses, vests, etc.), and flag equipment that has moved out of authorized zones.
Chad Aronson and Saeed Contractor gave a presentation on Uber’s journey with intelligent automation. The world’s largest ride-sharing company already uses technologies like named entity recognition (NER), natural language processing (NLP), natural language understanding (NLU), and artificial intelligence (AI) to automate customer interactions via chat and email.
SSON's State of the Shared Services & Outsourcing Industry Global Market Report