We discuss intelligent document processing and why it is quickly becoming an essential component of strategic business development.
Data forms the crux of business operations across many sectors. Data usually initiates most business processes in many functional areas like accounts payable, order fulfillment, customer service, medical claims processing, employee onboarding, etc. and even tech support. These are typical document-dependent processes that are mission critical to the overall functioning of departments, sectors, and the entire organization.
Mismanagement of such document-dependent processes can not just derail or stagnate entire jobs but also cost the employer significantly in terms of recorrecting and redoing tasks. The costs and inefficiencies associated with manual data entry and document processing lowers a company’s cjnahecs of moving on to the next stage of business development, and consequently achieving set business goals. A good way to mitigate these risks is by introducing intelligent document processing for streamlining mission critical back-office processes.
Intelligent document processing refers to the automatic acquisition of data from unstructured and semistructured data sources and processing it to be made ‘business-ready’ for assimilation and use by many business applications.
The data acquisition part involves automatic data capture and extraction from a myriad of diverse document types and formats, from different input channels including scan, fax, file transfers, email, MFPs, etc.
The data processing part involves performing automatic data validation by tallying the extracted data against databases, as well as verifying with assigned personnel for final approval, after which, the data is deemed fit for export to multiple business applications including ERPs and ECMs like Acumatica and NetSuite cloud ERPs, SAP Business One, Microsoft Dynamics 365, Sage, and Quickbooks . Approved data may also be reproduced in different file formats or systems of record including file storage repositories like Laserfiche and ODBC compatible databases.
Tight integration of document processing software with Robotic Process Automation (RPA) applications like Blue Prism and Automation Anywhere enables RPA workflows to access up-to-date approved data at any point in the workflows.
As is evident, intelligent document processing makes data immediately available for different use cases, making the technology indispensable for completing heavy document-dependent tasks.
The ‘intelligence’ in an intelligent document processing system comes from digital transformation technologies including AI, machine learning, intelligent data capture, OCR/ICR/OMR technology, robotic process automation, business process automaton, etc. These advanced automation technologies enable the creation of advanced data capture logic and intelligence-specific workflows. AI and machine learning are employed to ‘learn documents’ and understand each document format for easy data extraction.
To state an example, take the case of a company dealing with just 2 vendors. The invoices received from each of them may have a different format and structure — one invoice may have common invoice details like ‘TOTAL’ and ‘ITEMS’ in the body section of the document whereas the second invoice copy may have ‘TOTAL’ in the footer of the document. An intelligent process automation system with automatic document processing capabilities must be able to decipher the details of each invoice autonomously and apply a different data extraction logic to each. This requires cognition to learn each document format and apply a separate capture logic to each for optimum data extraction. Usually, most invoices have the sum total displayed at the bottom of a page, so an intelligent document processing system can build generic capture and data extraction definitions that are applicable for most invoice formats. It is when the invoice format changes that your standard ‘template-dependent’ capture logic fails — this is where machine learning and ‘cognition’ becomes necessary. For invoices that have formats and structures that are a ‘deviation from the standard’, generic capture definitions are rendered useless, and the system must have the capability to ‘learn’ the document and keystrokes performed to enter data from the document the first time, and apply that learning to process subsequently documents of a similar format. Machine learning is required for such ‘self-learning’ of new documents. Today’s intelligent document processing systems accomodate for document variability with advanced ‘self-learning’ mechanisms that help capture and process awide range of document formats.
As you can imagine, generic capture definitions are enough to extract data from structured documents. Structured documents are just that, structured, and can be captured using powerful OCR technology. Documents like standard government forms, academic question papers, or even most CRM data, all have a fixed structure with data organized in fields on the form. These types of documents can be regarded as images where OCR technology will be sufficient for data extraction. Unstructured and semi-structured documents like vendor invoices, customer sales orders, medical claims forms, emails, mobile data, sensor data, data from media files, chat and collaboration software, etc. do not have information organized in specific data fields, making it hard for traditional capture mechanism with solely OCR technology to extract the right data. OCR and many other traditional capture methods are template-based and rely heavily on context, making these technologies unsuitable for processing semi-structured or unstructured documents.
A combination of digital transformation technologies with cognitive capabilities of learning and deciphering are needed to accommodate a vast range of document types. Intelligent document processing does just that, achieving near-autonomous data capture and document processing, and adding to its built-in capture logic with every new document.
Data validation is an automated process and again employs digital transformation technologies to enable workflow automation of any number of complex document types. Validation is the process of checking extracted data against known entities and data sources to ascertain the document’s legitimacy. As an example, verifying vendor details including name, vendor ID (if there is one), address, date of purchase, etc. will help validate an invoice and accelerate payment approval for that particular vendor.
Workflows and logic for automatic data validation may be built from business rules and guidelines adhering to compliance and regulations of any kind. As mentioned before, data validity can also be established by tallying against known data sources like databases in ERPs and ECMs. For this, data validation techniques like cascading lookups or dynamic search can be employed. When tallying a medical claims form for validity, the intelligent document processing system can lookup a medical database for claimant details like name, age, and insurance type to make sure the reimbursement is being processed for the right claimant.
As was mentioned, approved data is exported in many different file formats as well as to business applications including ERPs and ECMs for business use. Tallied and verified invoices may be directly exported to an accounting software like Quickbooks or Sage ERP for finalizing payments to vendors.
It is this end use of approved and validated data that drives many business processes and helps complete final jobs in ERPs and ECMs. Today’s intelligent document processing systems offer tight API-level integration to many line-of-business applications, ensuring end-to-end processing of transaction and source documents.
A continuous data flow established right from the point of source when a document is captured to its final destination in a, say, ERP system ensures no data or functional silos. This end-to-end process automation framework due to prompt data management establishes what is now known as a connected enterprise.
Intelligent document processing helps remove data silos and establish a connected enterprise; and a connected enterprise in turn ensures no data leakages, inconsistencies, or data loss — creating a symbiotic business management model.
Artsyl leads the charge on intelligent document processing with its prepackaged, codefree vertical solutions designed to handle document-dependent processes like accounts payable, sales order processing, medical claims processing, and check remittance handling.
These ‘Action Solutions’ are built on top of Artsyl’s Intelligent Process Automation (IPA) platform, docAlpha for straight-through document processing. docAlpha is Artsyl’s flagship product and is the engine that drives intelligent process automation of many types of document-dependent processes.
docAlpha is a highly customizable intelligent process automation platform that can be configured to implement document processing for many different kinds of document-dependent processes. So, apart from the Action Solutions, Artsyl has configured for its clients intelligent document processing applications for processes in HR, for automating university admissions process, for automatic scoring, as well as records management and scheduling.
Customers have witnessed immediate benefits and returns from automating their mission critical process with the help of Artsyl including: