Raising An Ai Brainchild In The Mortgage Industry

Angel Ai takes on the next five years with a new ‘do and functions to boot

Angel Ai takes on the next five years with a new ‘do and functions to boot
Staff Writer

Pavan Agarwal likes saving money. When he looks around his office and sees only 11 underwriters at work, he smiles to himself. That smile is powered by Angel Ai. Let’s do some math:

Last year, Sun West Mortgage Company’s artificial intelligence platform, Angel Ai, processed 75% of all borrower-submitted documents. The remaining 25% – flagged for exceptions – had to be scoured by underwriters who manually submitted the data to Angel Ai.

This year, roughly 6.5% of borrower-submitted documents were flagged. In real numbers, 1,517 loans were decided by Angel Ai, and 107 loans were sent for a second signature.

There’s no carrot and no stick, but because Angel Ai learned, she was promoted, and Sun West CEO Pavan Agarwal was able to trim his underwriting team by 70% – saving money.

Watch it on The Interest: Five Years Of AI

Sun West’s Chief Operating Officer Jennifer Vallinayagam said they employed 40 underwriters at the peak in 2020. The 11 underwriters still on staff she calls “data scientists,” and they handle Angel Ai’s back-end functions. From loan approvals to restructuring options to reviews, they feed the problems they solve back into Angel Ai, which learns and improves as an AI-underwriting engine from these corrections being made.

But, building revolutionary technology does not come without these challenges. That’s what learning – and thus machine learning – is all about, improving over time. The power and opportunity of artificial intelligence is its ability to learn and adapt as the market continuously changes.

Though Angel Ai’s intelligence gaps must be overseen by the C-suite, Agarwal and Vallinayagram themselves, rapid change and rapid improvement define the artificial intelligence program. Agarwal’s ambitions for Angel Ai break beyond the mortgage business, all with the intention of driving engagement on the platform – and saving people money, be they bankers or babysitters.

Admittedly, the financial services industry has been slow to adopt customer service AI tools. But, adoption is expected to accelerate in the next 12 to 18 months, according to a new report published by Syntellis Performance Solutions.

“Financial institutions across the country continue to face numerous economic and market pressures in 2024, including interest rate shifts, evolving customer demands, and rising competition from neobanks and other digital-only providers,” said Eric Wheeler, senior director of product management at Syntellis Performance Solutions (now part of Strata Decision Technology). “Our survey shows that banks, credit unions, and other financial institutions are boosting investments in data and analytics, AI, and other technologies to enhance customer service, increase efficiencies, and streamline processes as they work to navigate these challenges.”

The survey offers a glimpse into the likely future for artificial intelligence in the financial services realm. As mortgage companies’ costs rise, AI solutions across the loan process, in customer service or underwriting, will save companies money and time by lowering headcounts and streamlining processes. Agarwal has been planning for this future for years, developing Angel Ai so that the program is ready when the market is ready.

But, Rome wasn’t built in a day, and neither was Agarwal’s easy smile. 

Overcoming Growing Pains 

Parvan Agarwal
Pavan Agarwal

Agarwal has been dutifully raising his brainchild, Angel Ai, since 2018, when the program was launched as an internal tool adopted from Sun West’s tech-oriented sister company, Celligence.

From mid-2018 to the third quarter of 2020, Angel Ai closed 40,000 to 50,000 loans. Through the end of 2023, that number moved north of 180,000 loans.

The program’s growth has accelerated since September 2022 when Angel Ai was launched beyond Sun West’s borders, meaning every licensed mortgage professional could submit loans to the platform even if they didn’t work with Sun West. From October to November 2022, six Angel Ai accounts were created. From October to November 2023, more than 5,000 accounts were created. 

> Pavan Agarwal, 

CEO of Sun West 

Mortgage Company

In total, more than 138,000 messages (inquiries about loans) have been submitted through the platform, with almost 23,000 unique conversations occurring in the last few months, says Vallinayagam. 

The platform attracts brokers like Greg Tanner of Carolina Home Loans. 

Tanner became an independent broker in 2022. While looking for loan investors to partner with, Tanner found Sun West and Angel Ai. “Now Sun West gets a large chunk of my business, and I’ve easily closed 20 to 25 loans using Angel Ai,” Tanner said. “I’ve been in the mortgage space for about 20 years, and the tool has especially helped me with income calculation, which is sometimes difficult, and it helps if you have a second set of eyes on it.”

He also appreciates Sun West’s “100% Assurance” policy. “Sun West stands by Angel Ai’s findings – even if it makes a mistake – and will still write the loan,” Tanner explains.

This sentiment shows that Angel Ai is still a work in progress despite its improving success rate. 

While Sun West’s promise to stand by Angel Ai’s outputs builds users’ confidence, it also means that the platform has to be spoon-fed data, babysat by its creators. But, underwriting and loan processing is only a sliver of the potential Agarwal sees for Angel Ai.

Big Plans And Blockchain

Although Sun West and Celligence have ambitious plans for Angel Ai, Agarwal only has a rough idea of how he wants to roll out all these changes as he continues to smooth out back-end functions.

“The first and most important is a full, chat-based loan application for the consumer so they can apply for a mortgage using just a chat,” the CEO affirmed. “We also want to implement a robust set of training data for consumers to ask any questions about financial services. Really, anything they’d go to Google for.”

The idea, Agarwal says, is that users can initially engage with Angel Ai by asking a general question, and then go deeper. This initiative would capitalize on a trend toward increasing consumers’ financial literacy in the mortgage industry. As a direct-to-consumer origination strategy, it also shows the potential for Angel Ai to disrupt other mortgage lenders’ businesses.

Agarwal, a proponent of increased adoption of blockchain in the mortgage industry, says adding NFT functionality ties these initiatives together. The Sun West team piloted the NFT function at a Diwali party hosted by Agarwal at his home in Puerto Rico. 

“When [you] click on that NFT, [you] can go and see it, see all the information,” said Agarwal. “An NFT is nothing more than a ticket. It’s a token … that is now with a record that the owner owns it. And the blockchain gives that ownership, immutability, and permanency.” 

Agarwal further explained how this works: “If and when [borrowers] give us enough data about themselves, we can say, ‘You qualify for $1 million. It says today, based on the information I have, you can close today on $1 million, so here’s a token.” With that token, which represents a pre-qualification for $1,000,000, the borrower can purchase a house. 

It’s a hefty promise that, once on the blockchain, can’t be taken back, which could lead to the borrower’s dream home – or a million-dollar mistake on Angel Ai’s behalf. But, that’s the bet Agarwal is willing to take.

“It’s a commitment to get the loan, right? There’s a handful of conditions on that, like the property you purchase needs to be eligible, the title has to be clear, and your financial position doesn’t change from the time we issue the token to the time we close the deal,” he said.

Given the blockchain learning curve and the confusion average consumers face in understanding the technology, Agarwal may face headwinds. Nevertheless, he’s trying to plan for a mortgage future where these technologies, he thinks, will be increasingly ubiquitous.

Differentiation Brings Consumers Back

Agarwal says the brand message for Angel Ai is ever-expanding, which is part-and-parcel of Agarwal’s user engagement strategy. The growth of the platform is tied to the roll-out of other features on Angel Ai’s drawing board – features that Agarwal banks on for expanding the brand’s audience inside and outside the housing industry.

“The future is that the app is opening up and becoming everyone’s personal assistant,” Agarwal said. “A lot of things we do today for [real estate agents] and loan officers can also be done for everyone else.” He knows it’s ambitious to try integrating every consumer outside of the financial services industry. 

It’s even more ambitious to hope they will use Angel Ai regularly. “I want to get it out in front of consumers so they can start using it … we already did the hardest financial service possible, which is mortgages.” He cites event management, landing page creation, and online invitations as non-mortgage use cases for Angel Ai.

“Whether you’re a loan officer hosting a first-time buyer seminar or a mom having a birthday party, you still have to do the same things,” Agarwal believes. 

Agarwal knows that Angel Ai is likened to ChatGPT. Thus, staking out the differences between these programs is a messaging and branding issue for Sun West to solve.

“We don’t want if a consumer has to ask why they should use Angel Ai, because then we messed up … we didn’t brand and communicate correctly,” Agarwal explained. “The question should be, ‘When do I use Angel Ai?’ not ‘Why should I?’.” He continues, “Angel Ai’s not going to do your homework … or generate your essay. But [it will] do other really interesting things that you may want and need.”

Some of those features do already exist. For example, the company created and patented “Ai-Sign,” which allowed them to do away with using DocuSign and replace it with an in-house product. 

According to Angel Ai’s website, Ai-Sign “provides the ability to slide through pages and view content before signing. At the end of each page, Angel Ai prompts users to add their signature by clicking on the designated option below.” Users can sign documents using their fingers, as well as save their signatures for future use on the same document.

Because Agarwal’s goal is to keep users on the platform as long as possible, the team is also working on a credit repair function. This is “the biggest thing on Angel Ai’s horizon” because half of all consumers need some form of credit repair, Agarwal says. 

“The idea is to give consumers a reason to stay on the platform and to be on the platform. And that’s why credit repair is so important; it gets them on the platform, gets us involved with them, and starts building trust so that when Angel Ai says, ‘Hey, have you thought about a mortgage?’ Or when they go to Angel Ai and say, ‘Hey, I need to think about a mortgage,’ we know their credit situation and can point out their best options.”

While differentiating from other AI programs remains a focus, Agarwal believes that he’s already beaten much of the competition in the mortgage space. Armed with its first-mover advantage, Angel Ai is backed by humans who know the program, and market, inside and out.

Tanner of Carolina Home Loans stands by this. “Human interaction is almost immediate if Angel Ai has an issue,” he said. “It’s helped me have confidence in my work, especially as a new broker.” 

> Pavan Agarwal

Challenging the Status Quo

Agarwal is confident that the next five years in Sun West’s growth in the artificial intelligence world will prove to stick with all ages – as he sees it, the next generation of homebuyers. He cites name recognition, brand swag, and being deemed cool as reasons why Angel Ai resonates with younger mortgage folks, and younger folks in general.

Take, for example, the “12-year-old girl test” that Agarwal swears by. Sun West’s marketing director Anthony Toro tells Agarwal that his daughter is the most popular girl at school because she has access to Angel Ai merchandise. This tells Agarwal that the brand will succeed with the general consumer market eventually. 

Angel Ai’s former alias, Morgan, failed this test. The rebranding of Morgan (to Angel Ai) during the middle of 2023 was intended to change people’s perception of the product. Now, people know that Angel Ai is an artificial intelligence system, unlike when “Morgan” was just seen as a woman’s name. 

But, what does a 12-year-old know – or care – about an artificial intelligence platform akin to ChatGPT? Agarwal says it doesn’t matter; what matters is brand stickiness. It’s a big risk to do a rebrand after five years of existence under one banner. According to Agarwal, Morgan was a hybrid of the phrases “Mortgage Can” and “Mortgage Magician.” 

It didn’t work.

“It’s got to go viral with children because they grow up to be young adults who will use your product. I’ve hit a home run with the brand as far as consumers appreciating and accepting the brand, people,” the CEO says. “Another good sign is [young people] love the merchandise … and they keep trying to get more. So I think we’ve hit a home run on the branding, and we’ve hit a home run in the messaging.”

Even though preteens have a long way to go until getting a mortgage, Agarwal says brand recognition should impact all ages. He drew parallels between successful brands like Nike, Apple, and Red Bull, highlighting the importance of a compelling message beyond product features. “Everyone knows Red Bull gives you wings,” Agarwal quips. 

What he knows, if all else fails, 12-year-old girls are keen on Angel Ai. 


To read more on this subject read Automation Attracts Investors Who Love Homogenous Loan Pools

This article was originally published in the Mortgage Banker Magazine May 2024 issue.
About the author
Staff Writer
Sarah Wolak is a staff writer at NMP.
Published on
May 02, 2024
Mortgage Banker Magazine
2024 Legends of Lending

Please join us as we salute the 2024 class of Legends of Lending

Mortgage Banker Magazine
Mortgage Banker Magazine
Assessing The Fed’s Heavy Hand In Mortgages

How far can the Fed take quantitative tightening?

Preetam Purohit
Mortgage Banker Magazine
Manufacturing Fair Lending

How data defines a modern theory of redlining

Ryan Kingsley
Mortgage Banker Magazine
Maturing With Mortgages

Homeowners want to age-in-place. Know the loans that help it happen.

Erica Drzewiecki
Mortgage Banker Magazine
A Watershed Moment For Trigger Leads

Pending legislation collars controversial data sharing practice

Bob Niemi
Mortgage Banker Magazine
Servicers Brace For Multiple Storms

Navigating Default Risk Amid Climate-Driven Disasters

Jane Mason

Become a subscriber.

Discover the story of your success. Subscribe to the nation’s longest running mortgage magazine.