The servicing of reverse mortgages is far more nuanced than conventional loans, with many moving regulatory parts that impact the way in which such loans are cared for by companies like Celink. With that in mind, what barriers might exist in implementing new technologies to simplify the servicing of such loans? Sergey Dyakin, chief information officer at Celink, in this session will explain how artificial intelligence is used in this space, what unique challenges HECM servicers face in adapting new technologies and what innovations he’s keeping an eye on going forward.
Maria Volkova (00:09):
Hello, my name is Maria Volkova and I’m the technology reporter at National Mortgage News. Today I am joined by Sergey Dyakin, he is the Chief Information Officer at Celink, a company that services reverse mortgages. Today we will briefly chat about how the HECM servicing space is embracing AI. Thank you so much for joining us today. Can you tell me a little bit about yourself and what you do at Celink?
Sergey Dyakin (00:39):
Absolutely. Thanks Maria, and thanks for having me here. As I mentioned, I’m Sergey Dyakin, I’m Chief Information Officer at Celink. We are the largest servicer of reverse mortgages and in Celink being CIO, I’m in charge of all things technology. In particular, the development of our proprietary servicing system, Reverse Serve, and things related to the borrower experience such as Borrower Portal.
Maria Volkova (01:08):
How is artificial intelligence currently being implemented and used in the reverse mortgage servicing space? And are there any distinct use cases of AI that you can tell us about in this space?
Sergey Dyakin (01:23):
Great question. I would preface it by saying that reverse mortgage industry has been traditionally slower in adopting new technology, and we’ve seen development in digital development in other tools. For example, automation would be slower than they would be in your normal mortgages or in other areas. But I think those things are slowly changing and we see examples of use cases for AI across the board. For example, we worked on building machine learning models to better understand the documents that we need in order to complete certain parts of the process of servicing the mortgages. If we didn’t have this tool developed, we would have to hire a small army of people to do this manually. Fortunately, machine learning models were able to crunch it in almost no time. So that’s one good example. At the same time, like I mentioned, I see things changing. I see that when I talk to the technology providers, they say they are now inundated by requests from everywhere, asking them to build a POC on AI. And so I would say this industry is not an exception in this regard. We ourselves decided to put more emphasis on the use of those tools. We sat down and identified several dozens, actually use cases related to better use of AI across the board. Things related to just helping our developers be better at writing the code or things such as helping the customer service to be more attentive, providing better information to the borrowers or getting other forms of information more efficiently to borrowers.
Maria Volkova (03:11):
So you mentioned that you guys are potentially implementing AI to better communicate with your consumers. Is that via chatbots?
Sergey Dyakin (03:21):
More so on helping the agents get the information so that they can answer on the phones more effectively. Keep in mind that the average age of our consumer is 72 years old, so it does take some effort for them to adopt to this technology. There is also another consideration to keep in mind, and I see it across other companies, not necessarily in the reverse mortgage industry. It’s not even necessarily mortgage industry. People are concerned about the consequences of simply unleashing the LLM models, which Chat GPT of the world into the wild because things might turn out not the way you expect. And we know the models hallucinate, we know the models provide the incorrect information, and so there is always a concern about that. Probably what we’ll see are the use cases that are internal facing as opposed to external facing to develop or take hold more effectively. Things related to helping us be more efficient, things related to providing this information to the agent so that they can actually communicate better to the consumers as opposed to necessarily getting something to the consumers directly. But I see those cases as well when you do communicate with consumers.
Maria Volkova (04:35):
Can you maybe tell us about some of the unique technology challenges in the reverse mortgage servicing space?
Sergey Dyakin (04:45):
Absolutely. As I mentioned, the demographics of our consumers is one of those challenges. Folks who are 72 and older, it takes some time to adapt to newer technology. Sometimes they may just be simply set in their ways in how they prefer to communicate. They prefer to call and have a conversation as opposed to go to the website and get their information. So that adoption does take time, but we see it, either with people who actually just recently got mortgages if you are 62, 20 years ago, we already had web and everything, so they were able to use the technology. They also maybe son or grandson or granddaughter helping those consumers. So we see that usage as well. The other challenge I would say is simply the smaller size of our industry compared to normal mortgages which exist dozens of millions, there is less than a million, less probably than a half a million of loans, reverse mortgage loans. So we have to invest knowing that we can only spread this investment over a smaller base of mortgages, so we need to be very careful about the return on investment when we put money into the development of new tools.
Maria Volkova (05:59):
And I guess going along with what you’re just saying, what are the impediments to putting in place technologies in the reverse mortgage servicing space?
Sergey Dyakin (06:10):
So some of them would be how we can get better adoption, whether from the borrowers or even in some cases through the agents, loan agents or other people who actually service the consumer because they also need to switch from one type of tool to another. That’s probably one of the things that comes to mind. The other one I would mention is mortgage industry is heavily regulated and so you need to be careful about what to produce. It has to go through multiple layers of analysis and approval just to make sure that we abide by all regulation and we are very careful about doing so. And so getting stuff out is not as easy as for your Instagrams of the world where they can experiment every time. We need to be very cautious about taking deliberate steps in getting technology out.
Maria Volkova (07:04):
Awesome. And to wrap this conversation up, one last question. What technologies are on the forefront that can have an impact on servicing as a whole or servicing in the reverse mortgage space specifically?
Sergey Dyakin (07:19):
I think that servicing remains a rather labor intensive business and document intensive business. So any technology that can rapidly automate this process would be very helpful. It’ll allow us to be scaled up or down depending on the circumstances, and allow us to handle situations in a more predictable manner. I’d also say that continuing development, the digital tools for the borrowers, regardless of the age, for that matter, they will adopt it eventually. It’ll help consumers more and more expect the tools to be conveniently available to them 24 7 available to them digitally. And our intention is to move from all the forms that they can do on paper to doing them electronically. So those are the things that I see that will be the focus in AI as we talked about this, AI can help quite a lot in those areas because it can improve the efficiency. It can help us get to the answers faster and it can help develop the deliver information to consumers on that relatively complex product more efficiently.
Maria Volkova (08:26):
Okay, wonderful. Sergey, thank you for taking the time and chatting with us today.
Sergey Dyakin (08:31):
Likewise, Maria. Thank you. Thanks.
