- 2 months ago
- #mortgageservicing
- #fintechai
- #voiceai
HousingStack, powered by HousingWire and The Basis Point, lets you quickly comparison-shop tech solutions critical to your operations today. Plus, the Basis Point’s real-time, on-stage analysis drills in on ROI and critical nuances of the technology to help lenders make decisions.
In this demo, Tavant’s Sundeep Mathur shows off a solution designed to streamline operations and increase efficiency for lenders of all sizes. From automation to intelligent workflows, this demo dives into how Tavant’s platform helps drive ROI while reducing operational friction — plus, get insights from The Basis Point’s strategic breakdown.
#MortgageServicing #FintechAI #VoiceAI
In this demo, Tavant’s Sundeep Mathur shows off a solution designed to streamline operations and increase efficiency for lenders of all sizes. From automation to intelligent workflows, this demo dives into how Tavant’s platform helps drive ROI while reducing operational friction — plus, get insights from The Basis Point’s strategic breakdown.
#MortgageServicing #FintechAI #VoiceAI
Category
🤖
TechTranscript
00:00Hello to everyone in the HousingWire community. I'm the BasisPoint founder, Julian Hebron, and welcome to HousingStack Live.
00:06This is a live demo collaboration between HousingWire and the BasisPoint launched at the Gathering Annual Conference this year,
00:12and it's where we take a look at key parts of the mortgage tech stack by category.
00:16Today, we're covering the state of mortgage servicing fintech with special guests, Tavant and Sandeep Mathur,
00:23and we're going to take a look at their voice AI that's powering multiple use cases in servicing.
00:28And before I tee up Mathur and Tavant's demo, let me just set the stage really quick.
00:35So servicing is the bigger opportunity between originations and servicing.
00:40Originations has been very reliable for 30 years at $2.02 trillion average fundings per year over those years,
00:47but at $14.5 trillion outstanding in mortgage servicing, it is the bigger opportunity,
00:53and it's where the most potential to modernize has been happening in these last couple of years.
01:00And it's critical because this is where you get all the time with the customer if we're doing this right,
01:07and if the software like Tavant helps build is powering us in the correct ways.
01:14You've got the borrower for between 1 and 12 months in a long lead purchase cycle on the origination side,
01:19and you can have them for life on the servicing side.
01:24If you're doing all of these things you see on the screen here, well, powered by software like Tavant.
01:32So customer care and hardship are the main things that we are going to focus on today with Tavant
01:38amongst all the things that you see here in the middle of the slide.
01:42These are the areas that have the most cost impact for servicers and the most upside for customer experience for borrowers.
01:54But let's just take a look at the cost.
01:57If you look at this pie chart from MBA, you can see that the top three costs are helping with customer service,
02:04helping with lossment, and of course, systems themselves.
02:06The basis points thesis here is that the system cost probably stays the same
02:13because these last few years have been about modernization of servicing systems,
02:17and only now, and especially with AI coming in to help with customer service,
02:24both in performing loans as well as customer service in the hardship scenarios,
02:30including loss mitigation, where that time for the servicers
02:35can be so burdensome and costly, this is where the costs start to come down.
02:41And as you're going to see from Sundeep's demo of voice AI and customer service for servicers,
02:49what happens, as I just said, is that the servicers can go way faster,
02:54and the consumers also just get a much better experience.
02:58And one of the things that I think you're going to see here is that multi-channel is so critical.
03:04The borrower can start anywhere.
03:05They can start on the web.
03:07They can start with chat.
03:08They can go over to voice AI.
03:11They can maybe run out of time that night when they're at home trying to figure out a question they have about servicing,
03:17and then they go over to their phone, and they can do it 24-7, and the voice AI will help them.
03:22Sundeep's going to walk us through all of that.
03:24So, Sundeep, I'm going to turn it over to you.
03:26You can take it away, and then we're going to come back,
03:29and we're going to run through some questions together.
03:31So, with that, let's have you show us voice AI and customer service for Tavon.
03:38Great.
03:39Thank you so much, Judy.
03:40We're so happy to be here.
03:42Really appreciate the time.
03:44So, today I wanted to just focus on a couple of things.
03:46Before we jump into the demo, I wanted to start with why are we doing this?
03:51AI technology is evolving so fast, and every industry is out there trying to figure out how do we leverage AI within our industries.
04:01And I think that as a lending industry, we also need to do the same thing.
04:05And one of the things that we're trying to do at Tavon is to help our customers figure out where can they get ROI by applying artificial intelligence.
04:14And, you know, the demo that you're going to do, and so we've kind of come up with a process to be able to help customers to do that.
04:20And one of the things that we've done today, what you're going to see today, the demo is all around servicing.
04:25And so what you're going to basically see is, you know, I'm the borrower talking directly to an agentic AI customer service rep who's going to help me through payments and refinance and going to help me through that process.
04:41So we built this proof of concept to basically help customers to see the art of the possible with artificial intelligence.
04:51And so this is just the beginning.
04:52We'll kind of walk you through a few different scenarios.
04:55But that's kind of what we're going to do today is this AI proof of concept that we've created.
05:01So let me go ahead and jump into the demo itself.
05:06Sorry.
05:07Okay.
05:09So basically what you have here is this is our customer service portal, and we have a few different scenarios.
05:18I'm going to jump into the first scenario here.
05:21And this is basically a borrower who has to make a payment.
05:26And so we're going to jump into the demo.
05:27Hey, Jerry, since your payment is due tomorrow, let's head over to the payment section to sort it out.
05:34Like I said, this is a typical customer service portal, servicing portal.
05:39On the left-hand side, you have a series of different options.
05:41The AI has recognized that this borrower has a payment coming up.
05:46And by looking at the servicing data, I know that.
05:49So the AIs realize that they need to go ahead and make a payment.
05:52And so this on the right-hand side is basically the AI, which we can either chat with it or we can talk to it in kind of a conversational mode, and we can turn it off.
06:02We can have it not talk to us or not.
06:04But I'm going to basically have a conversational interaction with it.
06:06Why is my payment $1,500?
06:11It seems like my payment has changed.
06:21It looks like your payment amount might have changed due to an escrow adjustment.
06:25This can happen when there are changes in your property taxes or insurance premiums.
06:29So as you can see, it's trying to, you know, trying to explain what the changes occurred and taking me over to the Documents Center where I could look at, you know, escrow balance or escrow reconciliation.
06:40But I have a fixed rate loan.
06:42Why would my payment change?
06:50Your monthly payment might change even with a fixed rate loan if there are adjustments to your escrow account.
06:55This usually happens due to changes in your property taxes or insurance premiums, which affect the escrow balance.
07:03Wait, it looks like my taxes went up by $100.
07:13So as it goes through, it's looking at all of the information that we've provided to it and responses that we've given to it.
07:21We've given it, fed it, a servicing policy.
07:23We fed it all the information around escrow analysis.
07:27So it's going to go ahead and respond back as to why I had a payment increase and that it was a tax increase.
07:39It looks like my taxes went up by $100.
07:41It looks like your property taxes have increased, which can affect your escrow balance and potentially your monthly payment.
07:52You can view the latest tax documents for more details.
07:56Well, how do I lower my taxes?
08:01So it's going to do the best it can to kind of help me through this.
08:03But one of the things I'm trying to show you is that as the borrower gets frustrated, one of the hooks that we built in, so it's going to give me an analysis of how I can go around lowering my taxes.
08:16But what I'm going to do in this is increasingly kind of.
08:19Lowering your property taxes can be a bit challenging.
08:22I'm going to shut off the voice for a second as we talk through this.
08:24So one of the things I'm going to show in this scenario is as the borrower gets more frustrated, we built in the ability to escalate this to a human.
08:31And so as I get more frustrated with the responses, I want to talk to a human, it'll ask all for me to escalate it to a human.
08:41Three, improve your property.
08:44Sometimes making certain improvements can qualify you for tax credits.
08:49Four, consult a tax professional.
08:52They might identify deductions or credits you haven't considered.
08:55So one other scenario that I wanted to show the team was the idea here is that as we recognize that this borrower, so that was a payment scenario.
09:09But if we recognize that the borrower is ready for a refinance or potentially a HELOC loan, the AI would recognize that and walk you through it.
09:17Hey David, welcome back.
09:19With your payment due in a few days, are you here to make a payment?
09:25No, I was interested in refinancing my loan.
09:28Hey David, I noticed you've shown interest in refinancing.
09:43Would you like to explore your refinancing options and see how much you could save?
09:48Yes, what information do you need?
09:50Okay.
09:58Great, let's get started with pre-qualification.
10:02Can you please provide your annual income, monthly debt, estimated property value, estimated credit score, and employment status?
10:12So as I go through this, I'll basically go to hit, yeah, my annual income is $95,000 and my debts are about, I have a $500 car payment, a $300 student loan payment, and my credit cards are about $1,000.
10:25So I grouped all that into one statement, but it's able to parse, the AI is able to parse exactly what I'm saying out, and I could have responded to more.
10:36Thanks for sharing your income and debts, David.
10:39Let's parse that thing out, figured out which part of it was my income, plopped it into the income piece, calculated by debts, by adding up those figures, and now it's going to walk me through the property value, credit score, employment.
10:49So it's just showing you that we're able to, in the oral conversation, we actually pluck information out of that, pre-fill these applications, and the AI is able to figure out taking you through this process step by step.
11:02And one of the things that we're adding into this is really this interaction ability for the AI to interact more closely with the borrower and to get them comfortable by using the technology this way.
11:13Yeah, and Sandeep, this is great.
11:17So I'm going to pick up where you left off on some of the Q&A.
11:20So leave it on this screen if you would.
11:22If I've got this right, it's a couple questions embedded in one to start us off.
11:28It's not just voice, it's chat, and it's happening both at the same time.
11:31So people can do chat or voice.
11:33And as you said in that last scenario, I was watching the main screen populate so the borrower could then take over.
11:43They don't have to use the chat bot.
11:45They could use voice or chat, or they can just say once these fields on the left in the main body are populated, they can just say, see my rates.
11:53Is that correct?
11:54That's right.
11:54That's absolutely right.
11:55So this is more of a transition state, right, where people are used to filling stuff into forms.
11:59So we wanted to be able to have the AI layer on top of that, kind of where things are evolving already are to a much more conversational thing, right, where you're interviewing the borrower, capturing that information in an oral conversation, and calling APIs in the back without the potential need for forms like this.
12:19But, you know, we want to get everybody comfortable with adopting these kind of technology.
12:23And so this is kind of the layering of AI on top of the current state.
12:28One of the things I love about this last use case is that it can be any of the above.
12:35And on a related follow-up question, I get jammed up right when I'm in the middle of this.
12:42I have to take a call, help my kid.
12:46I come back three hours later on my phone.
12:49Is all this stuff still going to be here?
12:52Absolutely, absolutely.
12:53So one of the key things is that, you know, in this conversation, we are basically going to summarize the conversation and keep both a summary as well as action items.
13:04So the next time the borrower calls, we kind of reload that.
13:07And if the system remembers the fact that you called three times, this is what we talked about, and this is where we are in the process.
13:14If you think about these agentic solutions, these agentic solutions have a goal.
13:18They have a mission in life.
13:20And they're trying to take you through this process.
13:23But, you know, you as the borrower, you can take it into any conversation.
13:27I ran out of time, but I can actually demo this thing and say, respond to me in Spanish.
13:32I can ask you any question about servicing, any question I want about lending, Freddie, Fannie.
13:36But it will kind of bring me back towards this, okay, we're trying to get you pre-approved.
13:42Take me through that process.
13:45And a couple more questions.
13:46We're getting a little close on time, but I want to do one quick one on what you talked about.
13:53I'm getting frustrated.
13:55I want to get to a human.
13:56I know we had to move to the next use case.
13:58Is that real time?
13:59Like, so if I'm ready then, right then, and I'm like, hey, my loan agent, Sandeep, I need to talk to Sandeep or somebody on the team.
14:09Am I being transferred at that time?
14:12So the way that we built this demo was that, you know, since we have your contact info, we built it such that, you know, the human would kind of call you back.
14:20And the system will provide you a security code.
14:23And then the human calls you back and says, okay, I'm calling from XYZ lender and my code is this.
14:30And so now you're able to know if the person calling you is in fact legit.
14:34But, yeah, absolutely.
14:35I mean, you know, they're, yeah, absolutely.
14:37You know, building in a real-time transfer is totally doable.
14:42And we do envision the applications would have a kind of a user side, right, an internal view of exactly what's happening here.
14:50And so that's also part of kind of what exists today.
14:55But, yeah, they're both being AI enabled.
14:58That's my last question, which is when borrowers, excuse me, when customer service reps are on their side and they're getting stumped, are you also developing AI to help them?
15:14Yeah, absolutely.
15:15I mean, you know, to my point in the beginning was that there is so much powerful usage of AI.
15:22We, as industry people, have to figure out where can we get that ROI.
15:27But there are tremendous opportunities.
15:28And so the most basic one that, frankly, in the last six months has just matured where, you know, it's sort of a straightforward, which is internal users using the knowledge bots to be able to answer questions about product pricing, policy, process.
15:46And, you know, we can tune these things to be able to respond based on the CSR's knowledge, right?
15:53If it's an experienced person, we can give you more bulleted stuff if you're a novice CSR.
15:58And we've also built this borrow simulator where we can give the borrow simulator different scenarios.
16:04And now they're basically testing the CSR.
16:07You're training the CSR on multitude of scenarios.
16:11So rather than, you know, you hired a new person, you put them through a few weeks of training, you put them on the phone, somebody's listening in.
16:18Now you can put them through a week's worth of hundreds of scenarios where the AI is pretending to be the borrower.
16:24And then there's one AI that's kind of coaching them.
16:28And now this person said, borrow simulator is another very powerful use case.
16:32But we don't want to sort of, you know, we're not trying to create something that everyone has to use.
16:36We're trying to find for different lenders where is the ROI, where are they going to find value in AI.
16:42And that's what we're kind of focused on is because different folks are different.
16:46And we want to apply the AI to where it's going to deliver ROI for our customer in a safe and compliant way.
16:53I think that's the second part of our sort of mission here is that in our highly regulated industry, you know, we have to be able to demonstrate no disparate impact and fairness.
17:03And the technology exists to be able to do those things.
17:06So anyway, leave that at that.
17:08And I think that Tevant has always been a leader, not just in building the tech, but as you say, building it specifically for the company.
17:17So everybody, Sandeep, Mathur, Tevant, demonstrating the latest in servicing voice AI and overall agentic AI.
17:28And we hope you come back and join Housing Wire in the Basis Point and do a Housing Stack Live follow-up because we know this stuff is evolving very, very quickly.
17:37Thanks, Sandeep.
17:38Thank you so much.
17:38I really appreciate it.
17:39And to close out today's Housing Stack Live session with Tevant, I just want to reiterate a couple of key points, which is that Tevant is building something in servicing that enables servicers to meet consumers where and when they are at any time of day or night on any platform.
17:59And using both voice AI as well as humans in the loop, this is something that we talk about with AI all the time.
18:07It's so much harder to do in servicing and Tevant has the big lender, big servicer experience to get this right.
18:15So big shout to Sandeep and team at Tevant for upping the game in mortgage servicing fintech.
18:22Thanks to Housing Wire for hosting Housing Stack Live with the Basis Point.
18:28And we're just so excited to drive the industry forward with these servicing innovations.
Recommended
1:09
|
Up next
9:14
1:56
5:24
2:12
1:01
17:44
17:19
1:01
20:36
0:58
1:10
Be the first to comment