Leading reverse mortgage lenders prepare to put AI in front of seniors

Top U.S. reverse mortgage lenders are entering a new phase in their artificial intelligence (AI) strategy: deploying customer-facing tools to engage more directly with seniors.

After achieving operational efficiencies by using AI internally, lenders now plan to roll out chatbots, voice agents and digital prequalification tools to a broader base of borrowers starting in the fourth quarter of 2025.

So far, their approach has been incremental — testing new technologies with employees first, then with wholesale and limited partners, and finally opening them up to seniors once feedback loops are in place.

“You start internally, like on things that have customer benefit but not direct customer impact, then you learn and expand out your reach,” said Brian Conneen, chief information officer at Finance of America (FOA).  

At FOA, employees now use a large language model trained on company policies and procedures to answer questions and link back to source documents. That shift has brought “a huge efficiency gain” while reducing escalations to managers, Conneen said.  

The lender isn’t yet comfortable putting GPT directly in front of customers. Instead, it is piloting AI tools with wholesale partners — considered more sophisticated users — to refine responses and guardrails. “Once we get that feedback, we could expand in a more restricted way to customers asking questions that we would monitor closely, like scripted responses,” Conneen added.

Meanwhile, FOA expects to introduce its AI-powered virtual call agent later this year. Seniors will be able to have natural conversations with an AI-powered assistant, which will hand off to a human loan officer if questions become too complex.

‘Hi, Bridget’

Longbridge Financial, owned by Ellington Financial, is also developing a voice agent that will allow stakeholders to interact with the company after hours, on weekends and at their convenience for routine tasks. But they will always have the option to connect with a human agent.

“We see these voice agents as providing capabilities,” said Bill Packer, chief operating officer at Longbridge. “When will we move these AI agents to be interacting directly with the customer? I think very quickly — probably, in the fourth quarter of this year, we will be ready on the servicing side to do some initial voice agent work.”

The company also plans to expand its AI chatbot, Bridget, which launched internally in July 2024. After a year of employee-only use, Bridget was extended to wholesale partners via the company’s portal this July. “We haven’t yet moved to having her face off with customers. We’re thinking about that,” Packer said.

Bridget has quickly become a high-traffic tool. In August alone, it handled 6,000 questions ranging from product guidelines to servicing issues and even responses related to natural disasters. It is now being trained to compare loan documents against investor guidelines. Underwriting is the most common area of inquiry — both from loan officers who occasionally originate reverse mortgages and from specialists in the segment.

Is the customer ready?

One challenge lies in the diversity of the reverse mortgage customer base. Borrowers can range from tech-comfortable 55-year-olds to seniors in their 70s who may prefer to avoid digital tools altogether. Companies must therefore design parallel experiences — both digital and call-center based — to meet customers where they are

“When we talk about older Americans, they are not a monolith,” Packer said. “Each of those populations across those cohorts has slightly different needs, views on it. If you don’t make it about the AI, but just make it about a seamless experience, then it never even enters their consciousness.” 

Longbridge reports that nearly 98% of its customers electronically sign origination documents, a sign that digital adoption can succeed when technology feels invisible. It comes back to making human-centered design decisions so that the technology is just a seamless background, Packer said.

Security and privacy remain key considerations. According to Conneen, collecting sensitive information, such as Social Security numbers, can create hesitation. To address this, lenders are experimenting with approaches such as requesting only the last four digits or deferring collection until later in the process.

“These are the kinds of things we’ll experiment with as we go forward,” Conneen said.

The AI journey

Reverse mortgage executives say AI is quickly gaining traction even as it remains in its early stages.

Longbridge has created cross-functional project groups in each core area of the business — marketing, retail, wholesale, servicing, default resolution, legal and compliance. Each group is tasked with exploring how to bring next-generation technology into its domain.

So far, the company has seen measurable productivity gains. In servicing, Longbridge deployed an exception system that eliminates “stare and compare” tasks. The AI reads documents, applies business rules and flags items for human review only when needed.

The result: three to five times higher productivity for the servicing team and up to a two-thirds reduction in errors. In some cases, the system even outperforms humans — particularly in deciphering handwriting.

Longbridge has already moved the tool into correspondent purchases and plans to fully automate that process once parallel testing with human reviewers is phased out. “Once we feel comfortable there, then it’ll continue to just move up that value chain forward into the rest of the origination space,” Packer said. 

Meanwhile, Conneen said FOA launched its first digital prequalification tool in June. The pilot started with the HomeSafe Second product before rolling out across multiple products as an opt-in option for customers. Currently, on any given day, about 60% to 80% of customers expressing interest have this option available.  

“As we continue in Q4 and beyond, we’re going to go deeper in the experience so that as they talk to the LO, they can continue to engage digitally on some things, or could come back.” 

That flexibility could help speed up closings. Conneen sees a 25% to 40% reduction in timelines as realistic once more customers engage digitally.

It’s not just that it makes the customer faster – as the customer self-services, it also makes LOs and customer service representatives more efficient. Less time spent per customer means each LO and CSR can serve more customers. “It becomes a virtuous cycle,” Conneen said.