PluggedIN: Intelligent Document Processing

Intelligent Document Processing transforms inert physical and digital documentation into dynamic informational assets. It is making its mark at present and is clearly poised to be an essential future business tool.

A QUARTERLY PERIODICAL

Q3 2021 EXELA TECHNOLOGIES

Intelligent Document Processing: The Present and Future of Automated Information Management Q3 2021

THE FULL COST OF PAPER

THE RISE OF INTELLIGENT DOCUMENT PROCESSING

Paper has been an essential business resource since its invention roughly 5,000 years ago. Its long tenure as a key human tool clearly shows its value, but it also suggests its impending obsolescence. As better technology is developed, and as the cost of managing paper continues to grow, some level of digital transformation is becoming a business imperative. Paper’s ubiquity may distract us from the full costs of paper-based business processes. The typical U.S. office worker uses more than 10,000 sheets of paper per year. 1 Given current paper prices, that’s about $80 per year, per employee. While that might not seem like a large expense, the true cost of paper extends far beyond the initial purchase. The full life cycle cost of paper has been estimated to be 31 times the cost of the paper itself! The main costs of paper documentation stem from its delivery, handling, storage, and retrieval, with copying costs at 33% and distribution costs at 56%. 1 We must also take into account the cost of real estate, storage accessories like folders and file cabinets, ink, additional staffing needed to manage paper, and, perhaps most importantly, the extra time spent handling paper. A single large filing cabinet can cost upwards of $1,000, and the square footage necessary to use onemight surprise you. The average filing cabinet actually uses 15.7 square feet of space! With the average annual cost per square foot of office space in the United States coming in at almost $39 (with a high of $85.82 in Manhattan), 3 a single file cabinet costs on average about $612 per year, just to store paper. Paper will continue to be a common component of business for the foreseeable future, but companies continue to find that a great deal of time, resources, and money can be saved by migrating to digital processes and through improvements in document management technologies.

As we enter the full-fledged digital age, paper isbecoming increasingly antiquated as better digital options emerge. Many companies are now thinking about digital transformation and developing strategies to achieve it. However, digital transformations are not instantaneous. The middle stages of this transformation are likely the longest, where a hybrid operating model that mixes paper and digital is used. Even companies that have been working toward complete overhauls for some time will likely find that a huge amount of information is still stored on paper. That paper-based information can be a hindrance, but it can also be turned into a major competitive advantage if companies know how to effectively utilize it. The massive growth of the document management market (digitization, imaging, indexing, and archiving) in recent years suggests that many are catching on. One measure put the value of this market at $34.21 billion in 2019, with projected growth to $57.56 billion by 2027. 4 Global trends on Google show the searched term ‘digital transformation’ moving from a score of around 25/100 in 2016 up to 100 in early 2020 and staying mostly elevated above 75 since then. In other words, we are now seeing peak search interest in that idea. 5

Projected Growth of Document Management Market

PaperCosts Money and Time 2

exelatech.com

Intelligent Document Processing: The Present and Future of Automated Information Management Q3 2021

Character Recognition Technology

95% of companies cite the need to manage unstructured data as a problem for their business.

A major driver of the recent wave of digital transformation has been advancements in character recognition technology. While the capability to scan documents into computer-rendered images has been around for a long time, an image on its own provides minimal value. Something must interpret the image in order to extract usable information, and until relatively recently, that work had to be done by humans. Data entry operations were, and in some cases still are, vital as a bridge between documentation and information management systems. Today, advanced optical character recognition (OCR) and intelligent character recognition (ICR) technologies enable scanned images of paper documents, or fully digital documents, to be automatically mined of their contents, generally with little to no human intervention. Whereas OCR technology can convert images with known fonts into machine-coded text, ICR takes that a step further by incorporating machine learning feedback loops for self-improvement over time. This enables ICR systems to more accurately process difficult content, such as novel fonts, images of poor quality, and handwriting.

Global Optical CharacterRecognition Market

Unstructured Data

By now, it’s common knowledge that effective use of data can yield incredible results for businesses of all types. Yet, despite the recent enthusiasm for mining and utilizing data, a significant portion of existing data remains out of reach because it lacks a known structure. Unstructured data is exactly what the name implies – data that lacks a predefined form by which software can identify and process it. A standard document type, such as an application form, can be created with determined fields that enable simple data extraction by basic OCR technology trained on a template. The software “knows” it is pulling a name from the ‘name’ field and an address from the ‘address’ field. However, an email, a web page, or a report has no such set structure. In order for machines to help us extract information from these unstructured media, smarter tools are needed. According to one study, 95% of companies cite the need to manage unstructured data as a problem for their business. 7 Most estimates suggest that 80-90% of all data is unstructured. 8 OCR and ICR offer the first step on the path to harnessing the power of unstructured data, and intelligent document processing (IDP) tools are pushing even farther.

10% Sturctured

80-90% Unstructured

exelatech.com

Intelligent Document Processing: The Present and Future of Automated Information Management Q3 2021

AUTOMATION AND DATA SCIENCE AS A SERVICE

While digitization, OCR, and ICR can effectively extract information from documents as never before, enabling us to begin to manage the unstructured data problem, they still have their limitations. But they do provide the groundwork for additional tools that can together create a more complete document processing solution. By adding natural language processing capabilities, cognitive computing, and robotic process automation (RPA) – and improving the machine learning capabilities of an ICR system – a complete

can be fully processed, analyzed, transformed into usable data, and then integrated into downstream systems – all without human intervention. The benefits of IDP to businesses in all industries are almost limitless. An invoice received via the mailroom or an email can be validated, matched to a purchase order, reconciled, converted into a payment order, and even paid, with minimal human intervention. An insurance claim can be received, processed, entered into the system of record, checked for errors, and approved for payment, with software managing

Intelligent Document Processing enables companies to unlock far more data than they had available in the past. When that data is added to the many other sources from which companies are pulling information, the amount available can be truly staggering. When it reaches a measure greater than what can be processed within typical data management systems, we call it “Big Data.” The total amount of data created and consumed in 2020 was estimated at 64.2 zettabytes, with projections that this number would reach upwards of 180 zettabytes by 2025. 11

document processing, content classification, data extraction and validation, and intelligent routing

a majority of the work. In a public sector full of standardized forms and

system can be created. Today, we refer to the work of this comprehensive tool set as Intelligent Document Processing (IDP).

structured portals, transformational outcomes are truly possible. Wait times and processing

times can be dramatically reduced along with long-term costs, all while public satisfaction

With natural language processing and cognitive

is raised substantially. Identify almost any type of business in existence

computing, meaning and context can be gleaned from text. With RPA,

today and IDP will fit somewhere into a plan for how

1 zettabyte 1,000,000,000,000 gb 1,000,000,000,000,000,000,000 bytes

information management, such as a transfer between systems, can be automated. And with improved machine learning, an IDP system can improve its accuracy over time and expand its useful reach into new areas, managing new media types. IDP systems take document management into the 21st century by leveraging the latest state-of-the-art technologies. As these systems continue to evolve, paper (and digital) documentation and other assets

to improve it.

IDP is still a fairly new technology class in a young industry, but it is growing fast. The global IDP market is currently about $0.8 billion in 2021, 9 but some estimates put its potential growth at as much as $12.45 billion by 2027. 10 A cursory glance at the trends combined with some modest assumptions about the evolution of these systems’ abilities suggests a very large future market by any estimation.

exelatech.com

Intelligent Document Processing: The Present and Future of Automated Information Management Q3 2021

The short answer? Most of them. Outsourcing data gathering and analysis can help companies of all sizes and in all industries better understand their customers, refine their products and services, and generate more powerful insights into the best ways to run their businesses. Working with a data science partner also accelerates improved data gathering methods and data analysis. Building a data collection and management system from scratch is a difficult process requiring a large investment in time and money. When data management is not a company’s specialty, developing these tools and procedures internally can be a major detractor from the company’s core mission. Even companies that are capable of managing andanalyzing their owndata can benefit from outside support, as well as the experience and computing power a DSaaS provider can offer. Partnering with a DSaaS provider also enables data volume balancing during periods of increased processing demand, and it can be a cost-effective way to leverage more advanced technology that your company may not otherwise have access to. Which Companies Can Benefit fromDSaaS?

Global Big Data and Business Analytics $274.3 billion by 2022

Most modern companies already know that data has become possibly the most valuable business commodity available; but many are still unsure of how best to leverage the data they have. Gathering more data is one thing. Solving the problem of how to utilize it is another. Companies that figure out the latter can create a significant differentiator and competitive advantage for themselves. Part of the trouble is that most businesses don’t specialize in the collection, management, refinement, and analysis of data - many don’t attempt to utilize data at all. Recent polling suggests that less than half of global businesses (48.5%) are using data to boost innovation . 12 Despite the known benefits data-driven strategies can provide, implementing data science into an established business process can be very difficult for companies without data specialists. One way these companies are now getting around this issue is by working with a partner that does specialize in data science. This type of service is now referred to as Data Science as a Service (DSaaS) or Analytics as a Service (AaaS).

48.5% global businesses are using data to boost innovation

$45.87 billion by 2026

$12.02 billion in 2020

Analytics as a Service Market

exelatech.com

Intelligent Document Processing: The Present and Future of Automated Information Management Q3 2021

IDP at Exela: Current Capabilities and Future Vision

Technology tools like OpenBox and Digital Mailroom are an important and valuable step along the path from paper-based work to fully digital solutions. When combined with other add-ons and additional services, efficiencies are greatly enhanced and valuable synergies become possible. For example, Digital Mailroom’s OCR and ICR technology enables features such as advanced auto-classification and routing, a check deposit workflow, PO-system integrations for streamlined invoice processing, auto-redaction capabilities for sensitive content, and more. It also connects with Exela’s DrySign e-signature platform for streamlined intake and execution of documents requiring a signature. Digital Mailroom and systems like it are essentially proto-Intelligent Document Processing systems. These systems offer huge value in today’s business environment, but the full complement of IDP technology components that have recently been developed offer even more – and the future potential of these solutions is truly enormous.

Exela has been offering digital transformation solutions since long before Intelligent Document Processing became a recognizable category of work. The roots of our technology trace back to the earliest days of image processing, beginning in the early 1980’s. Much has changed since then, other than Exela’s commitment to industry-leading document imaging through our IntelliScan suite of high-speed scanning platforms and our time-tested downstream information management solutions. The predecessors of the highly configurable, multi- use IntelliScan platforms were first implemented to process checks and remittance documentation. Today, IntelliScan has evolved into one of the world’s fastest high-volume scanning platforms capable of high-resolution scanning for nearly every document type imaginable.

The IntelliScan Raptor is a smaller "table-top" scanning system designed to be an affordable and highly versatile multi-function machine.

A Complete IDP Solution

While thescanningstepofdigital transformationmayonly turnapieceofpaper intoan image, it lays thegroundwork for the process of turning inert documents into dynamic digital assets . When combinedwith Exela’sOCR, ICR, MICR (Magnetic Ink Character Recognition), or Barcode Recognition technologies, the IntelliScan platform becomes a more complete document digitization and classification platform, with auto-sorting capabilities.

TheIDPtechnologythatExelaofferspicksuprightwhere Digital Mailroom leaves off to provide a comprehensive solutioncoveringphysical anddigital document intake, processing, classification, data extraction, validation, and enrichment, content indexing and summarization, and integration with downstream systems. Leveraging a growing model library of neural network classifiers that have, on average, been trained on 60,000 images, as well as machine learning-based parsers and validation modules, over 20 million transactions are currently being processed per month. Offered as an on-demand, cloud-hosted capture and processing service, or through a low- or no- code SaaS model (depending on the user’s level of sophistication), IDP can make a notable impact in all industries. At Exela, it is already providing significant value in the healthcare, insurance, legal, and public sectors. In one recent deployment, a major insurance company replaced their standard, rule-based OCR claims processing system with Exela’s scalable neural network-based document classification engine to achieve dramatic improvements in accuracy and processing speed.

The Next Steps

The next step in the chain from digitization to fully automated information management was Exela’s OpenBox software, which has been a valuable platform for our company for years. Capable of processing numerous media types frommultiple input channels, and built with full-featured automatic image import for more than 150 image formats, OpenBox was created to convert document packages into organized, indexed repositories. The enhanced document classification capability OpenBox provides is powered by a combination of classification engines, based on machine learning through various identification models. These models include: Self-Learning (training from example documents), Fuzzy Matching (probabilistic determination), Print Type (free-form classification), and Geometric (shape-based identification). With the addition of configurable business rules, cloud and local hosting options, and NIST and STIG-compliant data encryption, OpenBox offers a secure, flexible classification and information management platform. Similar technology forms the foundation of our highly successful Digital Mailroom (DMR) solution, which converts paper mail and other documentation into searchable, shareable, secure digital assets, accessible through the DMR Portal or by a customer’s preferred third-party platform integration.

Existing IDP workflow for live healthcare deployment

exelatech.com

Intelligent Document Processing: The Present and Future of Automated Information Management Q3 2021

IDP Results for a Major Insurance Company

On July 12, 2021, we announced that our robotic process automation platform, EON, has been successfully combined with our IDP technology to add additional attended and unattended automation capabilities into the system. EON now makes it possible to further remove the human element from the document processing system in areas such as data validation, correction, and transfer, which allows us to achieve even greater levels of efficiency and cost savings. In industries with large numbers of structured documents, such as healthcare, insurance, and the public sector, IDP systems have been making an impact within the most forward-thinking organizations. In areas where information is transmitted in various formats and through various media, where structure is generally lacking, IDP is poised to truly shine. The combination of intelligent language-processing neural networks, RPA, and related technologies is positioned to accelerate the next major sea change in the way business is done - and Exela finds itself out in front of this trend.

The Future of IDP

Exela’s IDP technology builds upon a long lineage of document processing systems that have been built to provide maximal automation in support of a global customer base. With use cases across numerous industries and document types as diverse as invoices, health records, insurance claims, and legal case files, flexibility and platformgenerality have long been a priority. When you can transformessentially any physical or digital document type into a manipulable data asset, major efficiency and productivity gains become possible. The evolution of IDP has grown from our efforts in the early days of the company and these new systems serve as a natural extension of our work in the business process automation space. The IDP industry today is still in its infancy, but we are at an important point along an evolutionary path that is converging with other related paths, all leading toward the type of true artificial intelligence we see in movies. Future generations of IDP technology are likely to produce systems capable of astonishing natural language processing accuracy and dynamic, contextual interpretation capability. We will eventually build systems that can read and literally understand what a document says. To date, this technology only exists in science fiction, but current work on IDP is setting the stage for that long-imagined future - much the same way that scanning platforms and OCR established the foundation that today makes IDP possible. Pushing the boundaries of accuracy and speed, and expanding the range of use cases, will continue to be goals within the IDP industry, and this holds true for Exela as well. As we deploy this technology in the service of our customers in numerous industries, we will continue to find ways to improve and broaden our capabilities. The future of work will be ever more digital and ever more automated, and we are preparing for that reality as we simultaneously help to bring it about. Building upon a long legacy as a global technology and service provider, Exela intends to remain at the forefront of innovation in the business process management industry.

Sources

1. 2.

https://www.informit.com/articles/article.aspx?p=1393497 https://www.industryanalysts.com/111015_konica/ vavv

3. https://www.commercialedge.com/blog/u-s-office-lease-rates-inch-upward-vacancy-still-in-double-digits/ 4. https://www.fortunebusinessinsights.com/document-management-services-market-102848 5. https://trends.google.com/trends/explore?date=today%205-y&q=digital%20transformation 6. https://www.grandviewresearch.com/industry-analysis/optical-character-recognition-market 7. https://www.forbes.com/sites/rkulkarni/2019/02/07/big-data-goes-big/?sh=50176f1a20d7 8. https://www.cio.com/article/3406806/ai-unleashes-the-power-of-unstructured-data.html 9.

https://www.reportlinker.com/p06093882/Intelligent-Document-Processing-Market-by-Component-Deployment-Mode-Organization-Size-Technology-Vertical-And-Region-Global-Forecast-to.html?utm_source=GNW 10. https://www.globenewswire.com/en/news-release/2020/12/02/2138036/0/en/Document-Analytics-Market-to-Reach-12-45-Billion-by-2027-With-COVID-19-Impact-at-a-CAGR-of-48-1-From-2020-Meticulous-Research-Analysis.html

11. https://www.statista.com/topics/1464/big-data/#:~:text=As%20of%202019%2C%20the%20adoption,the%20technology%20in%20the%20future. 12. https://www.statista.com/topics/1464/big-data/#:~:text=As%20of%202019%2C%20the%20adoption,the%20technology%20in%20the%20future.

13. https://www.mordorintelligence.com/industry-reports/global-analytics-as-a-service-market-industry 14. https://www.statista.com/statistics/551501/worldwide-big-data-business-analytics-revenue/

exelatech.com

Embracing complexity. Delivering simplicity. SM

exelatech.com 1 844-XELATEC

Page 1 Page 2-3 Page 4-5 Page 6-7 Page 8-9 Page 10-11 Page 12-13 Page 14