According to the IDC, over 80% of business information is trapped in documents of different types, formats, and origins – paper documents, email attachments, branch office scans, mobile device submissions, and portal uploads.
This huge amount of information, sometimes called “unstructured”, is not held in a digital format on databases, in-house software applications, third party systems, or spreadsheets. It is a point of difficulty for many businesses who need to analyze, monitor, manipulate, process, and control this information.
So, how can this complication be resolved? The solution is Intelligent Document Processing (IDP).
What is Intelligent Document Processing (IDP)?
Intelligent Document Processing is a software technology that automates the understanding of information contained in documents. It saves time, money and human effort while reducing errors caused by manual document processing.
Until IDP came along, a human had to look at unstructured information contained in documents and decide:
- What type of document is this? An invoice, an account application, a proof of identity?
- What information is contained in this document? PO number, applicant name, policy number?
- Who needs to process this? Finance, claims, accounts, or another team or individual?
IDP automatically recognizes the document type, extracts information, and releases the document and its information for use. Starting with a digital image of a document, IDP automates the tasks that previously required the cognitive ability of human workers.
How does IDP work?
Intelligent Document Processing learns to recognize and understand documents by using Artificial Intelligence (AI) technologies, including computer vision, Machine Learning (ML), and Natural Language Processing (NLP).
To kickstart its learning, IDP requires some initial training. For example, a human will feed IDP a small number of invoices (say 5 to 10) and tell the IDP technology, “These are invoices” and “Here is the invoice number”. In the language of AI, this is called “supervised learning”.
IDP then takes over and learns while working, becoming smarter with each new invoice it sees. This is called “unsupervised learning” and it’s where the revolutionary impact of AI proves true. The technology teaches itself, learns faster than a person ever could, never makes a “manual error”, and works 24/7 every single day of the year.
Benefits of Intelligent Document Processing (IDP)
Provoked by the disruptive effects of the COVID-19 pandemic on document-dependent work, many companies have applied IDP to automate mailrooms or to support remote working. This is because IDP transforms document-centric work. It can be applied in literally any process that needs to receive and understand documents, such as new account openings, customer onboarding, insurance claims processing, and underwriting.
Businesses that employ IDP are more responsive to their customers as previously slow, error-prone processes are transformed. The reduction of latency and cost enables business processes to be executed faster whilst suffering fewer exceptions.
What is the difference between OCR and IDP?
Optical Character Recognition (OCR) and Intelligent Document Processing are often confused as they are both used to achieve the same goal – understanding the content of documents.
OCR converts an image of text into readable text. When a potential customer submits a pay slip as a photographic image from a mobile phone to a bank, OCR converts the salary number on that pay slip into machine-readable text.
IDP is a larger technology that uses OCR but has more capabilities and does much more than just OCR. Interestingly, IDP will sometimes use more than one “OCR engine” to achieve better results. For example, one “OCR engine” may be good at converting printed text, while another may have particular success in converting handwritten text.
IDP surrounds OCR with additional capabilities that allow for document classification, finding and extracting specific information, validating its accuracy, and integrating data with other systems.
What is the difference between unstructured and semi-structured information?
Business information trapped in documents is sometimes called unstructured or semi-structured. This information is not organized consistently in a pre-defined manner and cannot be processed by automated systems.
A semi-structured document contains information you expect it to hold but you cannot be sure where that information will appear in the document. Every invoice, for example, will contain an invoice number, but different suppliers will put this number in different places on their invoices.
An unstructured document contains information, but you don’t know what information to expect. For example, customer correspondence will contain important points from your customer, or a legal document may contain details of a complaint. But not every customer correspondence addresses the same points and not every legal letter details the same complaint.
Intelligent Document Processing (IDP) can understand unstructured and semi-structured information, using various AI techniques, and pass that information to relevant teams and systems in a structured format.
How does IDP work with other Intelligent Automation?
Intelligent Document Processing (IDP) is used to understand the content of documents within many business processes, such as new account openings, claims processing and mortgage lending. Other technology is often required to automate the full end-to-end process involved. As a result, IDP is commonly deployed alongside other Intelligent Automation software. Examples of IDP working with other Intelligent Automation include:
- Robotic Process Automation (RPA): When data is released from a document by IDP, RPA software robots can automatically insert that data into a CRM, ERP or other in-house or 3rd party system. RPA can also help validate data released by IDP from a document, for example by accessing the customer’s account number from an in-house system and comparing it to the account number extracted by IDP from a new policy application document.
- Intelligent Process Automation (IPA): Some complex processes require more process automation than RPA alone can provide. To orchestrate workflow involving multiple people and systems, to drive processes so they meet Service Level Agreements (SLAs), and to handle escalations and exceptions well, IPA is often used. IPA can use the data released by IDP to impact the speed or direction of workflow. For example, the date on a new account application can trigger an SLA that allows 2 days as the maximum time allowed for the workflow to respond to the applicant.
- Case Management: Some processes, such as insurance claims processing, require IDP to gather multiple documents. Case management technology treats each claim as a separate ‘case’ and gathers all the IDP-processed documents and data together in a digital case folder. IDP can extract a Case ID or Claim # from incoming documents and automatically route the document into the correct case folder.
IDP and Ondox
At Ondox, we are experts in IDP and the other intelligent automation technologies mentioned above. We have helped customers in multiple industry sectors and business contexts choose the IDP that best suits their needs. We also help implement that IDP so our customers save time, money and human effort.
For most customers friction-free document processing requires more than just IDP. The Ondox digital mailroom solution is a cloud-based service that works alongside IDP to help complete the digital transformation of document workflows.
To learn more about IDP and how Ondox can help transform your whole organization, not just your mailroom, contact us. We’ll be happy to help.
