Mortgage processing is slow, inefficient and expensive. Everyone knows it. And everyone is trying to fix it. And?
After all this effort it remains slow, inefficient and expensive.
Clearly humans can’t solve this problem. So, it’s time to recruit the robots: “robotic agents,” that is, which use artificial intelligence (AI) to make human decisions at light speed. Granted, they can’t save the galaxy from invaders. But they can save money in mortgage processing—executing processes in days or even hours that once took weeks or months.
For lenders, robotic process automation (RPA) dramatically improve productivity and customer service, while slashing the total cost of mortgage delivery. For borrowers, the wait for weeks to get the verdict on their mortgage application is over. A rapid process improves their leverage in home buying negotiations and takes much of the stress out of mortgage acquisition.
Beyond pure speed, these agents also improve mortgage processing quality, accuracy transparency and profitability. Here are five ways AI-based processing can improve mortgages to make closing day less stressful for everyone.
One: Improved speed, speedy improvements
Robotic agents accelerate the mortgage process in several ways. First, they help employees accomplish day-to-day tasks faster and often take them on completely. They can key information into different systems and validate data via the “stare-and-compare method.”
It’s now also possible to reduce the time spent on mortgage document review. Depending on the state, a mortgage requires about 160 different documents, with any of them opened about 15 times during a loan review. Think of all the time and money people spend on this process—which could in theory require up to 2,400 total document reviews.
Agents also reduce or eliminate the need to re-key information into various parts of the system. They can scan and extract data from documents, recognize it and populate the system with the right information in the right place. This fax excels the time and cost of tackling silos of different products and systems that each need separate data entry.
Estimates by Realtor.com show that the entire mortgage process takes an average of 30 days in a normal market; this can go up to 45-60 days during high-volume months. This is because a good amount of “slack” is built into the overall process—meaning that in between steps, someone is waiting on information, or for a loan officer to review, approve or do something else.
AI-enabled robots gather, review and verify mortgage documents without the wait and with far greater accuracy. This creates an assembly line-like system that takes 24 to 48 hours to do its work.
Two: Fewer errors in a data era
Causes for mortgage errors abound. Missing or incorrect data can occur during the manual keying of information. For example, the income stated on a W-2 may not match the number on the application, or a street address could be incorrect or inconsistent.
This leads to delays in mortgage approval, frustration among homebuyers and added cost for lenders who must double back to fix the mistakes. These errors can also lead to compliance issues—and fines. Accuracy is crucial and AI technology can run on a near-constant basis without the human fatigue associated with lengthy and thorny document review.
Three: First-class compliance
Various state and local bodies have specific regulations that govern how information is gathered. Using robotic agents, a lender can pre-load a set of rules and regulations into the system to ensure the proper data handling. This can serve as a “pre-audit” early in the mortgage process and prevent compliance issues later on.
It can also raise red flags early for quick correction before they create bigger problems. AI-based systems can ensure that disclosures are handled properly and that the buyer has read them. It can provide a document checklist that further enhances compliance.
Four: Smaller paperwork piles
Imagine smaller piles of hardcopy paperwork coupled with much smoother document transfer, storage and retrieval. If a disclosure document must be sent to a buyer, a digital copy via email makes it easy to obtain an e-signature; that saves significant time versus waiting for physical documents via snail mail or setting up an in-person meeting.
While some mortgage documents still need hardcopies, they can often be scanned and uploaded to a digital system, making them part of digital workflows that are simple to monitor and report on. Losing a hardcopy mortgage document that might slow down an application becomes moot: Once it is digitized and “in the cloud” it will not be lost.
Five: Transparency that transcends
Robotic agents can not only extract data but also add a new level of context around it. For example, when entering the income of an individual, the system can determine how much of the earnings were from a regular salary versus a bonus, which could affect the loan decision. The right contextual data allows for much-needed proper analysis. Seeing the information in context means a full picture that leads to better decisions. A decision otherwise made in a vacuum could put the lender at risk.
This context also suits compliance purposes. If a down payment is made with a gift, it requires a verified gift letter and the system can identify whether it is missing, then request it. Such contextual intelligence provides actionable clarity upfront—and that greatly reduces conditions applied to a loan, saving precious underwriting time.
Parting shot: Mortgages move forward
Humans have gone as far as they can to improve the speed and efficiency of mortgage processing. It is time to relegate the truly robotic tasks to robotic processing. Having come of age, this technology can now take business processes to a new level of quantum improvements.
Lenders in the process gain new capabilities to delight customers, forge competitive advantage and drive profits. If it’s the vast expanse of mortgage industry we mean, maybe these robots can save the galaxy after all.
Want more Banking Strategies? Sign up for our free newsletter!
Arvind Jagannath is director of product management at AI Foundry.