Document indexing is the process of identifying and analyzing documents based on the information they contain. With manual processing, humans read through and categorize documents based on identifying features. With automated document indexing, this process is performed by AI programs and the identification is defined by a set of predetermined rules.
Loan processes have been protracted and complicated, with applicants needing to submit countless paper documents or even visit a loan officer in person to sign the necessary paperwork. Recently, more and more lenders have moved loan origination processes online, allowing them to be automated by AI cloud services. Unlike manual review, this automation happens far faster and with fewer errors allowing for an expedited loan approval process. Learn more about lending automation.
Intelligent Document Processing (IDP) is the ability to digitize documents and extract the data into a structured format that organizations can readily access. Businesses that rely on customer-uploaded or paper documents have a large data resource at their fingertips, but need an effective method to mine it. With IDP, they can structure data to gain insights into customers and their businesses. Read “5 Daily Tasks You Can Easily Automate with Intelligent Document Processing”.
Intelligent document processing (IDP) is the ability to process documents and simultaneously extract data into a usable format. It utilizes machine learning to simultaneously understand and boost accuracy over time. Optical character recognition (OCR) is a component of IDP that converts images of printed text into machine-encoded data. IDP utilizes OCR as part of document digitization.
Ocrolus incorporates OCR into our workflow, but it’s just one component. OCR struggles to get much above 85% accuracy across a representative mix of documents (it can do better at the “easy” documents, but not the tough ones). Fundamentally, our philosophy is that OCR will never be good enough for our clients to rely on it alone, and the whole point is for your underwriters and loan processors to actually trust the data they get back. That is why we’ve built both a machine learning layer, and included a “human in the loop”, to get us from 85% to 99%.
There are some things computers still aren’t very good at. That’s why we incorporated our “Human-in-the Loop” verification process into our broader workflow. Computers often still get stuck, especially on the “tough” documents, like cell phone images, scans, faxes, or just a document with a coffee stain. We built Human-in-the Loop document processing because our clients value a solution for all of their documents, not just the “easy” ones. They’d rather have Ocrolus’ human verifiers ensure data accuracy than make their own staff waste time doing it.
Over 100 lenders and fintechs have signed up for our intelligent document processing solution, including industry leaders like Paypal, OnDeck, SoFi, LendingClub, Plaid, and Square.
Ocrolus’ solution is not template-based. We use machine learning and AI to classify and extract data from documents like bank statements and pay stubs regardless of whether we’ve seen your form style before. Template-based approaches are inherently limited and so have trouble dealing with any irregularities, causing accuracy issues and often major delays.
Ocrolus processes millions of pages each week. Our model leverages each document to help make our process run smoother and our accuracy higher.
Currently Ocrolus processes English-language documents, however we are working to add new language capabilities in the near future.
Yes, Ocrolus users can file tickets any time of day and expect support around the clock.
Implementation timelines can vary based on the documents and integrations required for your solution. With the Ocrolus web app, users can start processing documents immediately. For our API users, an implementation timeline varies based on client resources, but most clients are in full production within a month.
Mortgage process automation happens through the digitization and transcription of borrower financial documents by automated software solutions, such as Ocrolus. It can also refer to the digitization of the mortgage handling process through the use of tools such as chatbots. Learn more about mortgage automation.
Traditionally, lenders have relied on metrics such as credit scores and pay stubs to evaluate borrowers’ creditworthiness. However, in an age when jobs increasingly do not meet the 9-5 norm, this often cannot provide an accurate view of a prospective buyer’s ability to qualify for a mortgage. Today, many lenders are looking for new ways to evaluate these non-traditional and self-employed borrowers and one of those options is through bank statements and cash flow analysis. When a mortgage is granted based on these other markers of creditworthiness, it is often referred to as a bank statement mortgage.
When it comes to evaluating non-traditional borrowers, it can be hard to discern the proper criteria by which to judge them as they do not comply with traditional metrics. Ocrolus helps lenders analyze other indicators of creditworthiness like cash balances and multi-source income. Ocrolus is able to provide insight into borrower cash flow dynamics, including income and revenue, expenses, transaction categories, ratios, and trends. With an understanding of cash flow, your team will be empowered to more accurately evaluate borrowers, both traditional and non-traditional alike.
Absolutely, our API is platform agnostic and we’re able to integrate with any origination system. Our implementation team has supported all of our clients through their onboarding and will work alongside your team to help integrate Ocrolus seamlessly into your current workflows.
Yes, Ocrolus has a formal partnership with ICE, giving us access to the Encompass LOS. Today we pull documents out of Encompass, run our classification, extraction, analytics, and fraud detection, place a .csv file in the borrower’s eFolder with the results. We are currently building additional API integrations that directly populate specific Encompass fields with borrower data. This will vary based on how you have set up Encompass and we’re happy to work with you to determine how best to connect with your system.
Generally yes, but more to the point your staff will be more productive as Ocrolus relieves data entry duties from business professionals. Ocrolus helps teams operate more efficiently, allowing each underwriter to handle more loan applications per day. This can offset hiring additional staff as companies grow or application volume fluctuates. We also help you get more out of your team by automating the most routine “Stare and Compare” type of work, triaging out applications that require scrutiny, letting your team focus on high-value work.
Bank statements can often be an accurate way to demonstrate proof of income. For many individuals who are self-employed or are employed in non-traditional ways, the usual methods of income verification, such as W2s or credit scores, are either unusable or inaccurate. Instead, a thorough bank statement analysis that measures cash flow over a period of time can accurately assess individuals’ income and average it out for the purposes of approving or denying a loan.
Mortgage fraud happens when a borrower intentionally misrepresents their financial information in order to fund, buy, or insure a mortgage. This commonly happens when borrowers alter their financial documents, such as bank statements and W2s, to show a different financial situation than is actually true.
There are two main types of mortgage fraud: fraud for profit and fraud for property. In fraud for profit, industry insiders use their inside knowledge to scam borrowers and institutions for a profit. In fraud for property, borrowers misrepresent themselves in order to gain or retain access to the property. For personal mortgage lenders, the latter issue is much more prominent and happens when borrowers falsify their financial documents or identities to gain access to more credit than they normally would be given.
Our approach involves a number of algorithmic and operational quality control checks to detect potential fraud. Ocrolus assesses submitted documents for signs of tampering and inconsistency, flagging edits that were made subsequent to the document’s creation and highlighting areas where changes may have occurred. Employees can then evaluate the documents to either confirm or further investigate potential fraud. Learn more about automated fraud detection.
Yes. By utilizing the power of AI, Ocrolus can help lenders perform income verification at scale in less time than it would take for manual review. Because Ocrolus can perform these verifications digitally and simultaneously, lenders have the ability to instantly throttle up or down depending on their loan volume.
Ocrolus processes most popular image formats (JPEG, PNG, BMP, GIF & TIFF), and can handle lower quality images, like those taken with cell phone cameras.
Currently, Ocrolus can process the following types of IDs for identity verification: US passport, US drivers licenses (any state), and non-driver state identification card (any state) and Social Security cards. We can also process verification of address documents like utility bills and voided checks.
Cash flow modeling provides a comprehensive analysis of a client’s assets, investments, debts, income, and spending and then projects those trends forward, predicting how cash flow will fluctuate given changes in income, inflation, and interest rates. These models can help lenders determine a borrower’s eligibility for a loan, especially if the borrower is non-traditionally employed.
Yes, Ocrolus can create custom metrics or analytics for you. To get started, reach out to your customer service representative and they will help you determine exactly what you need and how to set it up. Additionally, our income and revenue calculator is inherently customizable and easy to set up on your own based on the metrics your company uses to calculate these amounts.
Most borrowers interact with a variety of lenders in their finances; we have identified a library of lenders with over 700 variations on lender names. Because we work with a variety of vendors and process hundreds of thousands of documents daily, we’re able to add new aliases and update our library on a regular basis.
Yes. If an underwriter is looking to specify where data is delivered into their system, that can be done through our API custom mapping. In addition, our product offers the flexibility to call out transaction types within documents. For instance, we can group together recurring payments on a bank statement.
Pricing varies based on document type, extraction needs, volume, and contract terms. For a better idea of what your solution would cost, book your demo or reach out via our chat feature on the website.
Yes! By signing up for our free trial, you can analyze up to 100 pages of bank statements, W2s, 1040s and pay stubs. Sign up for your free trial here.
Blank pages are baked into our pricing, so while there is no extra charge, they are part of our overall service. For many lending applications, it is important to process all pages in a financial document, even blank ones as there is the potential for fraud. When processing blank or transaction-free pages, Ocrolus still evaluates the pages for signs that information has been removed or hidden. While it is possible for clients to remove blank pages, when clients have done this in the past, there is a high risk of human error and the wrong pages being removed.
Our pricing system is based on documents processed, not user access. Once the API is set up, you can have as many users needed at no additional cost; pricing will only change based on the amount of documents you process.