This is a BETA experience. You may opt-out by clicking here

More From Forbes

Edit Story

What Companies Need To Know To Begin Working With Intelligent Document Processing

Forbes Technology Council

Uday Birajdar, CEO and cofounder at AutomationEdge. Building delightful workplace experience with AI and Automation.

The paperless dream still remains just that: a dream. While digitization is catching on, companies are inundated by paper Know Your Customer (KYC) verifications, invoice processing, contracts, order forms, patient reports and insurance claims still being manually processed.

While document-processing technologies have been around for a long time, they have remained siloed in single-purpose customized applications, confined to a certain document type or a specific workflow like invoice processing. Documents received by healthcare providers, banks and insurance companies literally come in all shapes and sizes, from different vendors in different templates, letter-sized or legal-sized, with non-standard field signatures, rent orientations, with photographs stuck anywhere in the document.

Optical character recognition (OCR) struggles with these varied non-standard document formats and distortions. This has led to inaccuracies hampering straight-through processing rates, hitting productivity poorly. A standalone OCR is like taking a Polaroid. It captures just that singular, analog moment. It cannot be shared, and it cannot be copied, archived or saved to the cloud. It cannot be broken down into pieces or reassembled to tell new stories. The metadata, as it were, is completely lost at that moment.

One way in which companies are looking to capture that metadata and come one step closer to the paperless world is with OCR with intelligent automation, which can enable it to be used for end-to-end processing from the source of capture through recognition to classification, validation and verification to the integration with an ERP, EHR, core platforms, SCM or CMS.

This kind of system is known as intelligent document processing (IDP), and it has become useful as a way to process different kinds of documents (faxes, PDFs, handwritten documents, photocopies, images and grainy photographs) in an end-to-end manner. The system can recognize the field extraction of pertinent data (e.g., customer name, country, etc.) as well as classify by different categories (e.g., image-based, content-based, rule-based).

Verification is done manually, however, and IDP implementation can change the modus operandi of departments. That said, there are strategies organizations can take to help ensure a successful implementation.

How To Get Started With IDP

To get started with this option requires hands-on training of project managers and business analysts to understand IDP solutions. Employees taking part in the document classification may need to be retrained and learn how to assist in document verification, though it will open up their time for other, perhaps more useful, duties.

Because of these challenges, defining a clear governance structure is crucial for successfully implementing IDP. This means a structured change management and an alignment of IT and business functions.

Likewise, IDP initiatives will need to be aligned with broader automation initiatives, which requires comprehensive collaboration between the central data capture team and the automation center of excellence (COE). It might be helpful to embed the IDP team in the automation COE, as this can facilitate cross-skilling within the COE.

To measure and monitor the effectiveness of IDP solutions, KPIs will need to be proactively identified, defined and revised through the progress of the implementation. Providing robust training with historical documents, validating the correctness of the data labels to be used to train the model and setting up a well-structured central data management system with the availability of documents to train the model are some of the best practices in ensuring successful implementations of IDP.


Just like Polaroids have evolved to the force-multiplying benefits of digital photography, standalone OCR has evolved intelligent automation to enable end-to-end processing. As automation and digital systems sync up, we will gradually move away from the world of task-driven automation to intelligent automation.

As more and more technologies get added to the intelligent automation bucket, we will make way for systems to become more cognitive. These cognitive systems will help us achieve a paperless world and unlock the hidden value in our institutions, companies and governments.

Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?

Follow me on Twitter or LinkedInCheck out my website