As he attends conferences like the California MBA’s recent Mortgage Innovators Conference, Tavant’s Mohammad Rashid is looking for what’s happening “on the ground,” pockets of innovation he can take back and incorporate into the products and services the mortgage tech provider Tavant offers.
And pockets they are. In Rashid’s eyes, innovation isn’t happening broadly. And what has been happening hasn’t taken advantage of technological possibilities to the extent he thinks can and should be done. That’s a gap he believes Tavant can fulfill. And, in so doing, make for quicker, streamlined and more efficient and more frequent transactions.
“So, the industry is talking about innovation a lot, but it’s not really taken up technology to do that innovation,” Rashid, head of consumer lending practice at the Santa Clara-based company, told Mortgage Media’s Tom Wilkins at the conference. “If you take a look at AIML (artificial intelligence/machine learning), for example, it has made strides, large strides in other industries. But in terms of mortgage it’s starting to come in very slowly. There is an uptick, but it’s, again, slow.
“What we tried to do is basically come up with solutions that address — AIML techniques, for example, that address — classical mortgage problems. A simple example is predicting default, or predicting who’s going to refinance away. The typical way of solving that has been (to) put credit triggers or use statistical methods or heuristics, but AIML takes you much more ahead of that, right? And applying those AIML techniques actually gives you a big leg up. A lot of lenders haven’t really taken a look at it in depth, and we try to make it easy for them to understand, digest and accept that technology.”
Among those solutions: The Tavant VELOX digital lending platform, a suite of products, which Rashid said addresses pain points at all points of the transaction. Those include using AIML techniques to figure out who is likely to consider refinancing so loan officers can stay a step ahead, and give customers targeted, meaningful and contextual advice.
The Problem of Data
Tavant works with all players in the ecosystem (lenders, banks, title companies, mortgage insurance companies, appraisal companies, independent mortgage bankers), Rashid said. Lenders are the company’s “sweet spot,” he acknowledged, but all the spheres are interconnected, he added:
“The technology need, the innovation need is actually across the whole spectrum. It is not just the lenders. The lenders are kind of the center of the universe, but they depend on a mortgage insurance company to do their work. They depend on an appraisal company to do their work, the title company to do their work, the credit reporting agencies to do their work. So, the innovation has to really cover the whole gamut in order for the whole industry to come up.”
And, while the problems are different at each tier, the general issue is similar: How do I reduce the cost in the fulfillment process and make a profitable sale? And how do I eliminate the nonproductive, inefficient, costly and labor-intensive elements of the process?
That latter point is a big one in Rashid’s view: “I think the biggest problem today is with data,” he said. “So, in the mortgage industry you have a lot of people checking the checker, checking the checker, checking the checker. There’s a lot of manual filling in of data. It starts from the application process all the way through. And there’s multiple gateways and checkpoints that is basically checking the work that was done before. If we could eliminate that and we could have validated data and services up front that is taken all the way through and integrated and communicated with all the players in the ecosystem, it would remove huge amounts of redundancies and dependencies that we currently have.”
There is piecemeal improvement with the application of AIML, blockchain, robotic process automation and so forth — and certainly the digitization of data these days is an improvement over having to access paper bank statements and such — but the streamlining of data collection and retention needs to be more ambitious and encompassing, Rashid said, which won’t happen quickly.
“The overall innovation will really come when we have one validated source of truth, the data, that’s flowing across the whole board, across all the players. And for that to happen, you have to chip away at the problem one at a time. It is not going to change rapidly in the mortgage industry. There’s too many dependencies and too many players.”
Staying a Step Ahead
Rashid describes Tavant VELOX as “a suite of products that addresses pain points today that occur in the front and occur in the mid tier and in the back tier.” So there are actually different platforms – for consumers, loan officers, brokers, etc. that are designed to provide a better and more productive experience.
It addresses the front end by digitizing and validating data, eliminating a lot of repetitive and unnecessary hand-keying. On the back end, the FinConnect platform allows the user to connect to all third parties needed in manufacturing the loan — credit, title, MI and appraisal companies, day-one certainty vendors, and so on — for greater efficiency. (“One hop,” says Rashid.)
Then there’s the middle tier, the “fulfillment pipeline.” And there, he said, the site retention intelligence uses AIML techniques to figure out who’s going to refinance (maybe before they know themselves). That’s important in this climate: With interest rates the way they are, refinancing has grown immensely, more than predicted, and one may have more refinances than purchases. A professional needs to get a handle on this, needs to be ready so the customer will stay with them, needs to be proactive in giving them attractive offers. How to do this? Well, the traditional approach to predicting refinances is credit triggers, Rashid noted.
“But you’re putting a saddle on a horse that’s not only left the barn, it’s on a different farm altogether,” he added.
How do you stay a step ahead?
The problem: “It’s not just the signal that there is a subset of people who are going to refinance away, but I have to take them through the lead process, give it to my loan officers, and keep them. Give them offers that will keep them in my portfolio.”
Tavant’s solution: Apply AIML techniques to look at all attributes that come in from a borrower, property and loan — creating a deep-learning model, a neural network model, one that enables you to predict as much as three months ahead of time that a person or household is going to refinance.
This tech doesn’t eliminate the loan officer, Rashid reassured; the human professional is still essential to guide the buyer through the process. What it does (looking back to the front end) is take care of a lot of the more time-consuming labors involved with data while freeing the loan officer to concentrate on building the relationship. And back to the middle tier, the information gathered through AIML helps that relationship be more productive — the loan officer can be more useful to the buyer.
“What if I gave them tools that makes it easy for them to build that relationship faster and get them through the process faster and give them contextual advice, rather than generic advice?” Rashid said. That advice will likely be different for, say, a traditional home buyer or a millennial first-timer, he noted; the tools he speaks of gives the officer the means to give contextual, personalized advice — something Rashid says is lacking in the mortgage sphere though abundant elsewhere in the financial industry.
Not only does this potentially provide for a more fruitful result for buyer and officer, it could make the whole process much quicker, more efficient and less tedious, Rashid and Wilkins both noted — which could lead to people refinancing more often. Rashid doesn’t see why that decision shouldn’t be on the table at least once a year for households — maybe multiple times a year. But many people currently just don’t want to go through “that long, tedious process.”
“If I can give them the right product at the right time because the interest rates are the right place, their employment is at the right place, their DTI (debt to income) levels are at the right place — and be able to real-time notify them, ‘Here’s an offer, stay with me’ — this is where the matching of the borrower with the lender with the loan officer with the right product comes together, fulfills itself,” Rashid said.