AI-mortgage broker Ralo opens, raises $2.9M of capital

Artificial intelligence driven mortgage broker Ralo co-founders Arjun Lalwani and Helly Shah
Ralo's co-founders Arjun Lalwani and Helly Shah

Ralo, which claims to be the first artificial intelligence mortgage broker, has started business, having completed a $2.9 million seed round.

Processing Content

Its investors include Y Combinator, Manresa Ventures and Pack Ventures, along with angels such as Charles Ferguson (director of Inside Job, a documentary about the Great Financial Crisis) and Ryan Frazier (co-founder and CEO of Arrived).

The company says it uses artificial intelligence to automate the entire process, saving consumers money by cutting out loan officers, processors and underwriters, which it claims take hefty commissions.

Its founders, former Google Product Manager Arjun Lalwani and former Google Software Engineer Helly Shah, have become licensed loan officers and are currently Ralo's only employees. Shah also formerly was a quantitative analyst at Goldman Sachs.

Can AI originate a mortgage loan?

A recent Mortgage Bankers Association white paper said the Secure and Fair Enforcement of Mortgage Licensing Act does not set requirements for AI. But other laws, including Truth-in-Lending, require a person with a Nationwide Multistate Licensing System number to sign an application.

Meanwhile, slightly more than half of the prospective homebuyers participating in a Veterans United Home Loans survey said they would be comfortable making their purchase without any human involvement.

Just under 90% would share personal financial information with a lender's AI tool in exchange for real estate finance advice, while 68% said they trust the mortgage information technology provided.

What led to Ralo's creation

Lalwani said he applied for a mortgage a few years ago in San Francisco, and "the experience was very outdated."

The process was trying, including a denial and having to restart with another bank.

Similarly, Shah looked to get a mortgage in New York. "She faced the same problems where she did not get favorable terms, and was unaware of how to negotiate and shop," Lalwani said. "We both were at Google and we were talking about just all the cool AI tech we're building and how getting a product on Amazon is so easy, but for the largest purchase of our life, it somehow feels very outdated, and feels like we're being transported back to the early 2000s." Much of the process still involves sending .pdf files back and forth to the lender, he said.

They decided to focus on fixing the inefficiencies in the process.

"We have built an extremely smart harness on top of open source LLMs," Shah said, referring to large language models.

Ralo doesn't have its own models. "By leveraging the technology that's already out there, we have automated 1,000s of details to make this work," she continued. "It has taken us about a year to build, and it gets better every day as we process more and more loans."

Even its processing function is fully automated, without the need for a human in the loop to manage documents.

"We felt that a lot of the tech we were building could be used to improve this transaction for the end consumer," Lalwani said.

How the company is doing so far

The company's first loans closed in March. It is currently licensed in California, Colorado and Texas, with plans to expand to other states in the future. In the next six months-to-one year, it plans to be licensed in 10 states.

It is averaging 17 days to close a loan right now, 15 for a refinance and around 20 for a purchase, Lalwani said. 

From the few loans Ralo has done so far, they have learned every customer's story is different and each file is different, Shah said.

"The main problem that we have been trying to solve is, how do we automate these 1,000s of details," Shah elaborated. "So as we process more and more loans, the next customer we have, their experience is even better, because our automation has been able to capture so many nuances that every borrower's file faces."

While many borrowers have plain vanilla files, several, especially in the non-agency side of the business, might not have all the documentation a conforming loan has.

Ralo's next challenge is to capture these nuances and complexities to ensure it can bring the same experience to all borrowers, she said. 

Why Ralo looks to stay as a mortgage broker

Ralo also plans to stay as a mortgage broker, noting this gives it the ability to shop for its customers.

If the borrower needs to speak with a live person, Lalwani and Shah are available to talk on the phone.

But many consumers today don't necessarily look to speak with a live person for a growing number of purchases they make, not just consumer goods, but even for stocks, Lalwani pointed out.

Mortgage regulators should understand the modern consumer's needs and design lending laws accordingly to fit, he said.

Several wholesalers have approved Ralo as a broker, and part of the behind-the-scenes work the company does involves making sure its systems are compatible with the different technologies and workflows lenders those lenders use, Lalwani said.


For reprint and licensing requests for this article, click here.
Artificial Intelligence Originations Underwriting Mortgage technology Marketing MARKETING TO BORROWERS
MORE FROM NATIONAL MORTGAGE NEWS
Load More