We can all remember the days when document capture revolutionized the enterprise with the promise of the paperless office. Walls of file cabinets would disappear, mailrooms would shrink, and desktops would become visible again. Digitizing documents came from two approaches: in the mailroom for incoming correspondence and when staff completed a transaction with a customer and sent the file to be archived. The basement often served as a hub for industrial scanners and copiers where thousands of documents would be scanned and stored in a data warehouse. Capture was a driver for cost and time savings, but it locked important information away from ever being valuable. Not anymore. A new generation of intelligent document processing solutions have catapulted documents, in all its forms, to be the impetus to the entire customer experience journey.
Let’s look at applying for a mortgage, for example. Prospective home buyers would meet with a loan officer to complete a paper-based loan application. That application, along with hard copies of documents that showed proof of being able to repay the loan, such as W2s, paystubs, bank statements, and other trailing documents, would sit on the loan officer’s desk for months until the loan closed, when then the file would be digitally archived. Copies of all the documents would be digitally scanned and sent to other partners within the loan cycle ecosystem where they often would print the documents for their own files.
Within the last five years, this process has become completely digital and mobile where consumers can apply for a mortgage from their smartphone, including snapping photos of various documents and sending it directly to their online file, without ever needing to meet a loan officer in person. Better yet, all the information can be easily accessible to anyone within the chain of custody authorized to access it. This same scenario is now also common with Departments of Motor Vehicles processing REAL IDs; insurance policies and claims, whether commercial or private or for life, auto, or health; within the healthcare industry for medical records among various providers; and within supply chain and logistics where suppliers and transport carriers all need access to the same information.
The impact to the customer? A smoother experience that is faster, more efficient, and convenient.
The digitally transformed process all begins with documents and drives how the customer experience will go – whether information is missing, needs to be verified, or false. But we need to expand our concept of documents. Documents are more than Microsoft Word files. They can be invoices, emails, PDFs, images, photos, driver’s licenses, passports, and other documents that may come into an organization from various channels such as mobile devices, online conversations, fax, scanner or email – all in complex forms and sizes that contain various fields and could be ad hoc. However, simply digitizing documents still leaves too much reliance on manual involvement to transfer data from digitized versions of documents to enterprise systems, which can lead to errors, delays, and frustrations to customers.
Trained document models delivered as skills
The advances in artificial intelligence (AI) has led to a new generation of capture known as intelligent document processing (IDP). It leverages AI to do more than digitize documents — it learns like humans to recognize certain data, understand it and reason what to do with the information no matter which format the data is in. Everest Group defines IDP as software that captures data from documents, then categorizes and extracts relevant data for further processing using machine learning, natural language processing, optical character recognition, and computer vision. It is a strategic initiative for enterprises that recognize documents are at the center of the customer journey and want to not only achieve cost savings, but also improve their workforce productivity and employee and customer experience.
IDP’s ability to transform data has made it a complementary, yet necessary, tool to all intelligent automation platforms for enabling them to process data. Robotic process automation (RPA) platforms especially benefit from IDP because its software robots are unable to recognize and process the unstructured data that makes up 80% of enterprise content. While RPA has become a popular tool within the enterprise to replace manual data entry, it has many challenges for accurately and efficiently recognizing what to do with data. IDP gives bots the brains to process content, and in effect, deliver content intelligence to the organization.
This was made evident during the Small Business Administration’s Paycheck Protection Program, which offered small businesses COVID-19 relief during 2020. Many U.S. banks turned to RPA to help process millions of loan applications. They used IDP with RPA to find the relevant information from applicants’ trailing documents to process applications 30 times faster than they were processed just weeks before when they had a team of staff at their desks. This enabled banks to approve loans faster with accurate information and save thousands of small businesses from shuttering, thus allowing their employees to keep their jobs.
The agnostic nature of advanced IDP platforms makes it ideal for use with other intelligent automation platforms like SAP and other ERP applications, CRMs, and BPMs.
Low-code/no-code and cognitive skills trend
Traditional capture always required a high level of expertise that involved studying for certifications and painfully training software to train documents, then integrating it within enterprise systems. Advanced IDP platforms with its OCR, ML and NLP capabilities took the pain out of training any document type and API connectors made deployment easy, but it still required working knowledge of ML and legacy capture. Fortunately, dynamics in the tech landscape have transformed the accessibility and useability of IDP.
The reported shortage of tech talent has led many organizations to upskill their current workforce with new digital skills, and has workers eager to learn new capabilities that have made them become more comfortable having their daily work augmented by digital colleagues. At the same time, the democratization of AI into enterprise software has resulted in the emergence of low-code/no-code (LCNC) platforms that knowledge workers can easily control and has extended into the IDP market.
It’s not necessarily the LCNC platform that workers care about. They’re used to learning to work with different enterprise applications. And whether they install and manage it themselves as citizen developers or have IT staff do it for them, it’s the new level of cognitive skills they now have access to that makes the difference. New generation IDP no-code platforms don’t require machine learning experts or prior experience working with legacy capture, but deliver the same power and breadth of capabilities organizations need to keep content intelligence and engagement at the center of the customer experience.
The cognitive skills available to workers are specific abilities to identify data from various document types. A pre-trained cognitive skill injected into a document process includes digitizing the text in a document, from any channel of input like email, fax, mobile, and scanners, and interpreting the objects contained in the document. It could also be trained to understand specific document types, regardless of the variations, so that the skill can first identify a document in a process and then intelligently locate, extract, and validate the data. It’s as easy as dragging-and-dropping a skill from a digital marketplace into the no-code IDP platform to design and train a document skill at design time and it will continuously learn over time as document variations are processed.
Within our mortgage example, when the Center of Excellence team is ready to automate processes, a loan estimate or loan application skill could be used to train the platform to process those specific documents. In finance and accounting, an accounts payable analyst could access an invoice or purchase order skill to quickly automate AP processes.
The use of cognitive skills as part of an IDP no-code platform is a new way of thinking about document-centric processes. It’s very similar to how a skill is added to an Amazon Alexa device to automate the lights in the living room. However, document processes are more complex than an Alexa device and their workflows impact several other systems. To avoid automating a broken process or causing havoc to another team, it’s important to understand process workflows before jumping into a project and know how workers interact with them. Creating a digital twin of how your processes work is a smart way to analyze desktop user interaction data with process details mined from system event data to gain insight, oversight and foresight of processes. This process intelligence also provides the proof you need that the document processing automation project you are choosing will deliver the expected impact.
Overall, enterprises have made great strides in creating the paperless office. New advances in customer engagement like conversational AI, messaging apps, chatbots and even video will further complicate document processing and compliance and auditing standards. The goal is continuously having the ability to quickly leverage content intelligence, and to deliver the optimal customer experience and know that, as you consider no-code IDP platforms on your digital transformation roadmap, you’ll be able to find a skill for any document within your organization.