Nation’s biggest mortgage lender, UWM, jumping on AI bandwagon

By Matt Carter

Although it doesn’t work directly with consumers, this year UWM has rolled out AI tools to help mortgage brokers serve both homebuyers and homeowners who might benefit from refinancing

From lenders to loan servicers, the mortgage industry is embracing AI, and the nation’s largest provider of home loans is no exception — even though it’s a wholesaler that works with mortgage brokers and not directly with consumers.

Introduced in May, United Wholesale Mortgage’s ChatUWM is a “smart search” tool that helps mortgage brokers understand lender guidelines, matrices and tools and technologies that UWM makes available to brokers.

ChatUWM taps into UWM’s knowledge base, The Source, pointing brokers to full articles or pages it’s pulling information from so they have additional context. UWM announced new features on Wednesday that allow users to query documents and explore loan options for borrowers based on their financial situation.

“Brokers can now upload any PDF and engage in dynamic conversations with the documents,” UWM said in an announcement. “Whether extracting critical details from an appraisal or clarifying seller credits from a purchase agreement, ChatUWM simplifies the process, enabling brokers to quickly understand key information.”

ChatUWM relieves brokers of chores like calculating borrower income by allowing the tool to extract the necessary data and do the calculations when they upload documents like W2s and self-employed income statements.

Armed with a credit report and W2, ChatUWM can even analyze the documents, ask follow-up questions needed to fill in any gaps, and recommend loans that would be good fits for the borrower — along with the broker’s compensation for each option, UWM says.

UWM’s pivot from refi to purchase loans

Source: UWM earnings reports.

With purchase loans constituting 81 percent of UWM’s Q2 2024 mortgage originations — more than double the 40 percent share registered during Q2 2021 — helping mortgage brokers serve homebuyers is crucial.

ChatUWM lets brokers upload a purchase agreement and ask, “What are the seller credits?” and “Is the buyer contributing anything additional?”

But refinancing boomed when mortgage rates plummeted during the pandemic, and on UWM’s Q2 earnings call in August, CEO Mat Ishbia said he saw refinancing as the biggest immediate opportunity if mortgage rates continued to fall.

The Pontiac, Michigan-based wholesaler will put AI to work to grow that business, too, having rolled out a new tool last month that sends pre-validated refinance alerts to borrowers who look like they would be able to reduce their monthly payments.

“When a refi market hits, one of the biggest challenges loan originators face is staying in front of their past clients and competing with the substantial budgets and aggressive marketing tactics of large banks and retail lenders,” UWM said in announcing the new AI-powered KEEP tool on Sept. 11.

According to UWM, KEEP constantly monitors and validates borrower data points on previously closed loans. When it identifies a good refi candidate, it automatically sends the homeowner an email explaining what they might save by refinancing and provides a link to a loan application.

“KEEP’s proprietary technology will create the easiest refi process ever — providing opportunity to the broker and significant savings to the borrower,” Ishbia said in a statement. “It gives brokers the competitive edge they need to succeed in this market and further solidifies why the wholesale channel is the best place for consumers to get a mortgage and loan originators to work.”

Many of UWM’s rivals — including Rocket Mortgage, which UWM surpassed in 2022 when the pandemic-fueled refi boom faded — have also been scrambling to incorporate AI into everything from marketing to processing and underwriting loan applications and selling the loans they originate.

Rocket, which has said its investments in AI have cut turn times by 25 percent, says that when lending picks up again, it will be able to rapidly scale its business without having to go on a hiring binge.

Home equity line of credit (HELOC) provider Figure says it’s developed a process powered by OpenAI’s GPT that’s cut manual upfront document review labor by 93 percent while raising customer satisfaction scores.

Loan servicing giant Mr. Cooper, which collects monthly payments on more than $1 trillion in mortgages, spends several hundred million dollars a year on call center operations and expects at least $50 million in annual savings from its investment in a multiyear AI project.

While not all companies have the resources to develop AI tools in-house, a growing number of tech providers are offering to help.

Mortgage tech provider TidalWave’s AI-powered residential mortgage engine, SOLO, is now integrated with Fannie Mae’s Desktop Underwriter and Freddie Mac’s Loan Product Advisor.

Simplist Technologies offers Sonar, an AI-powered solution integrating loan origination software, point of sale software, a product pricing engine (PPE) and customer relationship management.

Mortgage technology provider Maxwell offers an AI-powered business intelligence tool for lenders that allows them to make database queries in plain English.

Mortgage capital markets technology provider Polly is embedding AI within its PPE, enabling tools like an “interactive copilot for loan officers” the company says is capable of processing, interpreting, and recommending “a range of outcome-driven results.”

More than FOMO

A June survey by intelligent automation company ABBYY found that “FOMO” — the fear of missing out — is a key driver of AI adoption.

In a breakout of the survey focused more tightly on the banking and financial services sector that ABBYY provided to Inman, 69 percent of IT decision-makers acknowledged that they fear being left behind if they don’t adopt AI. However, 64 percent said customers expect it, and 63 percent said they are looking at AI as a way to increase their company’s efficiency and provide customer service.

Two-thirds of IT decision makers in the financial services sector (66 percent) said their companies are already using generative AI tools like ChatGPT, large language models (LLMs), chatbots and digital assistants. More than half of companies in the sector (56 percent) said they were using purpose-built AI tools such as intelligent document processing (IDP).

“It’s interesting to see the financial services sector’s faith in LLMs despite the initial skepticism towards their tendency to hallucinate or provide inaccurate results,” ABBYY executive Maxime Vermeir said, in a statement. “This indicates that the market is maturing by incorporating purpose-built tools like IDP into their AI strategy, using them to address specific business needs and enable more trust in LLM-powered solutions.”

The survey also found that AI adoption rates varied widely among departments within financial services companies:

• Finance administration (48 percent)
• Marketing (46 percent)
• Development and engineering (46 percent)
• Customer service (44 percent)
• Sales (40 percent)
• Operations (38 percent)
• Procure to pay (31 percent)
• Accounts payable (28 percent)

The biggest concerns about using AI among IT decision-makers in the financial services sector were the cost of implementation and technical complexity (35 percent), followed by data requirements for training (34 percent) and potential misuse by employees (33 percent).

While 82 percent said they trusted AI, the biggest concern among those who didn’t was the reliability and accuracy of data (50 percent), with 42 percent citing worries about cybersecurity and data breaches, AI bias and data privacy.

The Original Press Release