How to Operationalize AI in Collections: Boost ROI with Virtual Agents | Ep. 223

Adam Parks (00:08)

Hello everybody, Adam Parks here with an episode of Receivables Podcast. Today I'm here with a good friend, Mr. Shantanu who's joining us from Prodigal. Today we're going to talk about artificial intelligence and how artificial intelligence can help us in the debt collection industry to move from a whole pile of behavioral data to actionable insights and actually being able to turn and operationalize some of these

 

Shantanu (00:27)

Thank

 

Thank

 

Adam Parks (00:34)

great pieces of information when you're able to put the puzzle together and start to see what it actually looks like. So, Shantanu, thank you for joining me today and sharing your insights. I really do appreciate it.

 

Shantanu (00:44)

Thank you for having me. This is a really well produced and thoughtful series that I'm excited to be part of.

 

Adam Parks (00:51)

Well, I really do appreciate you participating in it. For anyone who has not met you at a conference and you've been around on the conference circuit for more than a few years now, but for anyone who hasn't had the pleasure of getting to know you personally like I have, can you tell everyone a little bit about yourself and how you got to the seat that you're in today?

 

Shantanu (01:08)

Absolutely, would love to. I'm the CEO of Prodigal and we are an intelligence layer for our industry. This intelligence layer is used by human agents and virtual agents. I got to this point by essentially being at the intersection of the confluence of technology and finance for about two decades now. I grew up as a computer science engineer. That's where my education was. But very quickly, been kind of migrated over and

 

was doing finance for pretty much all of my working life until I realized that the financial services industry as a whole is changing dramatically with the advent of technology. So for the last, like say, 12 years or so, I have been working at the intersection of finance and using technology to make it better. For the first handful of years, for the first three years or so, I led data science and analytics at a lender itself. The lender was called Fundbox, fantastic team.

 

I was at Fundbox where I got a first-hand view of the collection industry and the loan servicing space broadly. And inspired by what I saw, I realized I really needed to do something about some of the opportunities that technology could solve and address. And with that in mind, we started what we're doing right now. And this has been a phenomenal journey. I got a chance to meet some great people like you, Adam, and many others who have dreamt of technology being

 

such a force multiplier in our industry and we are happy to play a little bit of role in that transformation.

 

Adam Parks (02:36)

Well, it sounds like you've had a storied career coming into it. That background has really played out well as Prodigal has become kind of a named brand across the debt collection industry. Now, I know you're rolling out all kinds of new products and it's been a year or two since you and I have really had a chance to sit down and kind of talk about what's been happening over at Prodigal. Tell me a little bit about some of the products that you've rolled out recently at your organization.

 

Shantanu (02:56)

Thank

 

Absolutely. We keep our ears very close to the ground on two fronts. One is what is happening on the customer side, which is everyone who's a collection agency or third party and early out and so on and so forth. And then the other place that we keep very, very close watch on is technology. And not only is there a change in regulation, not only is the change in kind of administrative policy and not to mention the macroeconomics factors like unemployment that are impacting our customers, which are

 

collection agencies. Coincidentally, at the same time, we have a massive wave of AI leading changes in our industry. And that is something that we are extremely excited about. And using AI to completely transform the operations and then the P&L essentially of servicing and a collection team either within a lender or a third party collection shop is something that we just realize is inevitable. And so with

 

That in mind, we essentially expanded our intelligence layer to not only make sure that we are monitoring and solving a lot of the QA problems, we made sure that we are not only making agents more productive on every single interactions and more efficient between them, but we extended it to say that everything that we know about a consumer, we should understand those behaviors, map those personas, and eventually evolved into a pro agent, which is essentially a way for

 

agencies to do outreach and tackle inbound phone calls that come in from consumers. And that expansion is very logical for us and it builds upon the best of everything that we've learned in the industry so far enabled by the latest and greatest technology that is coming out of the Silicon Valley every day.

 

Adam Parks (04:29)

 

Well, you guys are doing some really interesting things. you know, one of of the as we were preparing for this podcast today, one of the things that we were talking about was the application of artificial intelligence from a voice and conversational perspective. And one of the things you had been sharing with me was about how you had gone from kind of assisting the agent on the phone and providing them with insights and real time information to actually using that to create a virtual agent itself.

 

So it seems like a much more natural progression for your organization. Now, everybody out there tells me they got AI. I can't tell you how many messages I get on LinkedIn every week from everybody who's got artificial intelligence. But it's your approach, I think, to the challenges of the debt collection industry that really have differentiated your ability to operationalize this type of technology. I you guys were a first mover across the board with the ProNotes and...

 

Shantanu (05:12)

Yep.

 

Yeah.

 

Adam Parks (05:38)

even ProAssist and some of the other technology

 

Shantanu (05:39)

Yeah.

 

Adam Parks (05:40)

that you guys have developed that help the workflows to continue forward. But you had said something interesting in your introduction here today talking about the intelligence layer. And that's where I really want to dig in today and try and understand because there's a mountain of data sitting within our CRMs. And how are we supposed to take this behavioral information

 

Shantanu (05:45)

Yeah. Yeah.

 

Yeah.

 

Adam Parks (06:02)

and operationalize it. Like how do we turn this raw data into something that actually drives an insight and drives an action that moves us towards the resolution of an account.

 

Shantanu (06:09)

Thank

 

Very true. In fact, I was talking to a friend of mine, Carl, from Professional Credit a couple of weeks back, and he said that at some point, people used to write account information on, you know, recipe cards and stuff. CRMs essentially that, yeah, CRMs essentially became a digital version of that recipe card. And it took your file cabinets and digitized it. That in itself doesn't guarantee you any intelligence. And that is where we kind of came in and said, hey,

 

Adam Parks (06:24)

literally.

 

Shantanu (06:41)

intelligence is something that sits on top of it and probably also is really plays an important job in contextualizing everything that happens in that recipe card. So it is not just a file cabinet, it is a lot more than that. And for us, that journey makes the most natural sense. AI is a means to an end. Coincidentally, we did start right around a month of where this paper called Attention is All You Need, which which led to this current wave of AI.

 

in like late 2017. So we started right around that time as a company and the beauty of it is we kind of knew that a big change was afoot. We ourselves couldn't actually precisely pinpoint what that would look like. That being said, we were so much paying attention to everything that was going on that we kind of stayed on top of it, launched a lot of products that were very much cutting edge at every point along the way. ProAgent is again,

 

like the next logical iteration of it where the intelligence layer not only looks at your CRM, not only contextualizes every single file and make sure that the consumer gets the service they deserve and expect, but you're able to actually do it at a fraction of the cost. And so for us, it is less about AI as a end onto itself, AI as a means to an end so that you're able to actually turn all the data, all the...

 

all the information that you have written on recipe cards in your CRM and turn that into intelligence and hence revenue. And that is the kind of intelligence where we build. We call it the Prodigal Intelligence Engine or AKA PIE, where every single thing that we build, every single application that we build sits on top of this PIE. So essentially our product line, and that is exactly how we think every industry is gonna go, but there is an intelligence layer, PIE, that we really invest in understanding

 

every single minute detail about the consumer that is useful. And then there's a series of apps that sit on the layer on top of PIE. Let's think of it as layer cake that both build upon all the intelligence that's present in PIE, the Prodigal Intelligence Engine, and more importantly, in turn, kind of help make PIE better. That also helps us, this intelligence layer also helps us cross-pollinate across applications. For example, in a number of cases, an agent, a person might make a

 

contact with an account, but they are unable to speak at that time, they're just busy. We're able to capture all of this through ProNotes aided by ProAssist. If they're unable to complete that information, that kind of again goes back to our intelligence layer, so that if they were to call back like, know, after dinner, as an inbound call, we understand that they kind of considered some offers, they went through an RPC process.

 

Shantanu (09:29)

and they're really looking to make a payment. And we treat them very, very differently because we have a very clear note in our Prodigal Intelligence Engine, PIE, that explains this. And as a result of it, a pro agent, which is a virtual agent, acts not as a standalone entity, but it acts in conjunction with a lot of the intelligence that we have on that account at that point. And that just makes it very easy to deploy, very effective when it comes to, you

 

Shantanu (09:59)

helping the consumer where they need help and very cost effective.

 

Adam Parks (10:04)

So it's interesting when you start talking about the virtual agent side. It's in the 2024 TransUnion Debt Collection Industry Report, when we were looking at the deployment of artificial intelligence across the space or across the respondents, one of the things that we saw was the, I want to say it was a 16 % of all the companies that were using any kind of artificial intelligence were even considering doing any kind of voice. Now,

 

Shantanu (10:19)

 

Yeah.

 

Adam Parks (10:29)

I think that also is partially because in 2023, 11 % of companies said that they were even willing to explore artificial intelligence. In 2024, 18%, a 40 % increase. Now granted, 18 % is not a massive number, but that was a question that I asked last year. I'm really curious to see what that response rate looks like in 2025. But the conversational piece, I think is the...

 

Shantanu (10:37)

Yeah.

 

Right.

 

Thank

 

Thank you.

 

Adam Parks (10:56)

that scares people the most because they think about generative AI, they think about ChatGPT, and they think about all of the hallucinations and the stories that they've heard about the hallucinations that come from that type of technology. You know, from your perspective with that data intelligence level, what kind of guardrails are in place to avoid that type of risk in the future?

 

Shantanu (11:09)

Thank you.

 

100 % again that is a very important question even more so in our industry than someone you know trying to build a telemarketing bot and trying to you know migrate it over to our industry it is for very right reasons something that everyone should be mindful about we have our agents our virtual agents are actually effectively like PhDs and doctorates in all of the things that are relevant

 

Adam Parks (11:28)

Sure.

 

Shantanu (11:41)

with respect to the law and regulations, even at a state by state level. So certain states are treat servicemen differently from other states and our agents are actually taught all of those things. And then in turn, they also actually have very, very clear set of guidelines that they do not cross. the way, so AI doesn't mean everything is non-deterministic. There are parts that are clearly well-defined, well-understood, and you have to make sure that your system stays within that.

 

Shantanu (12:09)

Also

 

the reason why just taking something off the shelf and applying it really nearly into your shop is very likely a bad idea because you don't know what all engineering and guardrails need to be put in place to give you the kind of experience that is both productive but also legal. And we do a lot of that ourselves in multiple steps. Essentially it involves three steps. One is kind of training or

 

where you look at all of the information, what all the agents need to do. Second is kind of fine tuning. Fine tuning is another set of guardrails that make sure that the agents are, so first kind of just, the training gives them a set of knowledge. Fine tuning gives them guardrails and make sure that their performance is within those guardrails and you set up those fine tuning guardrails really, really strongly. And prompt engineering is the last mile thing that in turn,

 

constantly and independently keeps checking for the performance to make sure that it's within the guidance. And having these like multiple levels of checks give you a certain level of performance that you can't unless if you just copy something on the shelf. That requires months of concerted efforts. That requires availability of a lot of information in pie that tells you like, this is how you treat customers in New York state, which is different to say the least.

 

Shantanu (13:27)

And so all of those things kind of go into making and giving an experience that is again compliant, but more importantly, like effective. Otherwise you're just not going to see the value of the product. The other thing we should all remember is like machines in some ways are actually find it way easier to memorize and recall information than humans, especially once it gets past like 50, 100 documents. And that again is an ability that we are able to unlock because machines are able to kind of give you the

 

best perfect answer in many ways and they don't tire, they were ready for seven to give an experience that is way better. But I must say that none of this would have been possible unless ProAgent was able to stand on the shoulders of all of the giant apps that came before it, know, ProAssist, ProNotes, ProInsight. Because in some ways, our account level intelligence by product intelligence engine is able to really think about

 

what went on, making sure that they were being helpful and then ProAgent then is able to extend that as well. So I think this information that you mentioned, adoption speeds will pick up. have a great deal of thoughts on like how do we make implementation and adoption and success quick, somewhat readable and impactful, but you can't do it as a standalone exercise.

 

Adam Parks (14:39)

 

It's part of a greater ecosystem, right? Being able to do these things and bring these things together has to be part of a greater ecosystem, which I think is an interesting approach from what you're saying about going from pro assist and into pro agent and kind of that, that called growth curve. Now, how do you go? How do you go through you're working with a number of different clients in the space, a lot of our mutual friends? How do you get them comfortable with the idea of using generative

 

Shantanu (14:49)

It does.

 

Adam Parks (15:17)

voice. I feel like that's like that a genetic voice is is like the fear factor. And like, how are you? What testing or methodologies are you using to help get them more comfortable with the idea of using this type of technology?

 

Shantanu (15:32)

So synthetic voice again is something that we test out a lot and while it is machine generated we have come really long way as a company as a civilization in generating synthetic voice in the last like even six months and as a result of it we are able to kind of make progress and give comfort on very to two very very clear dimensions. One is performance and performance measured purely in the context of latency.

 

It requires a lot of engineering to generate relevant stuff, relevant pauses and respond in a way that feels human-like and doesn't feel like you are praying to some machine gods and they may or may not respond in three to ten seconds. We take in like there's a lot of like micro detailed engineering that goes into making sure that our latencies are extremely human-like. You also don't want to be too quick.

 

Shantanu (16:27)

because then that is not how humans work. So there is a sweet spot. There's a Goldilocks spot between being too quick, unnaturally quick and being so slow that people are going to drop off. And we show it in performance, both statistical as well as anecdotal to our customers saying, look at this machine and look how well engineered it is so that you feel like it is very responsive. It is like, you know, steering a car really, right? You don't want

 

like a lot of our cars have power steering you do not want to be it to be like overly sensitive where it is like you know starts wobbling at a touch of the moment you touch the steering wheel you want it to give a certain level of comfort and responsiveness and that is the kind of performance that we deliver. And then the second is kind of more around accuracy and the tone and timbre of the voice. So, we want to make sure that machines do not sound robotic obviously, they will never say that they are not machines they will always communicate that.

 

Shantanu (17:25)

Even then you want to make sure they are comforting and they are natural and a lot of sound engineering goes into ensuring that. So the first set of things is a lot of product engineering to drive latency down. The second is a lot of sound engineering to kind of make the quality very human like and both of those we show to our customers through statistical performance where they are able to test a lot of these things for like days and weeks and it goes into

 

Shantanu (17:52)

just anecdotally that they are able to experience it themselves. Everyone has a little bit of an aha moment when they see a product and they're like, I always knew this was possible, but this is the first time I'm experiencing it. And once they experience it, they obviously bring in more friends, their colleagues, and they beat it up even. They really like pressure test the whole thing and end of it, they get comfortable.

 

Adam Parks (18:16)

It's interesting, you know, we released an article earlier today about the eight personas of consumers in debt collection. And it was something that I had been working on and trying to think my way through. as I was writing it, I started to realize that there's a whole subset of consumers and it's not a small subset anymore that are really looking for the self service technology.

 

Shantanu (18:35)

It is.

 

Adam Parks (18:36)

especially if you're talking about consumers that took out their loan online, their expectation is that they're going to be able to service it online. And I think the even when you're talking about major banks, the online banking experience that they had when they were with Bank of America or Chase or whoever. Is there expectation going forward in the debt collection space? And so what is that going to start to look like from a consumer standpoint? Now, the consumers seem to be more open to the idea of

 

Shantanu (18:52)

.

 

Adam Parks (19:04)

using artificial intelligence in a number of different ways. I found myself doing it with my airline not that long ago where I realized that I can sit here and wait for 12 minutes for a live person or I can just ask these questions to this bot right here and come up with a resolution. And I took that path. It feels like consumers are just getting more comfortable with the idea of communicating with a

 

Shantanu (19:07)

Yeah.

 

Adam Parks (19:29)

computer versus communicating with a live human because 90 % of those online and online interactions are happening

 

Shantanu (19:36)

Yep.

 

Adam Parks (19:36)

out of country anyway. And so being able to talk to a voice that you're kind of finding yourself in a new world of a back and forth where you can actually get a response from it instead of me just yelling agent, agent, agent at the top of my lungs every time I'm on the phone, you can feel that change over the last maybe 12 to 18 months.

 

Shantanu (19:58)

For sure.

 

Adam Parks (19:59)

And as we look at the technology curve itself, like how quickly

 

technology is being adopted and how fast it's actually moving forward, it feels like things are moving a lot faster these days than they ever have before. Things that we would look at and say, it's gonna be a two-year growth curve, or we might have that product in three years. It's like we might have that project in like six months, three months. And it's starting to get a little crazy. Has that been, what's your experience been like as a...

 

Shantanu (20:04)

Yeah.

 

Adam Parks (20:26)

as a creator of technology in a world that's now moving at a lightning speed.

 

Shantanu (20:33)

If you have the right tools, if you have the right people and you have built things in a way that's modular, speed of development has

 

become really, really high and we are all for it. It is phenomenal. Our industry also has seen a mind shift shift mindset shift from leaders and executives where they understand that something new is happening and they need to be prepared for it. And on the flip side, think consumers, kind of referenced, consumer trends have changed, especially on three dimensions. The first is convenience. You kind of reference that you don't want to be

 

waiting for a human agent to be free and you would rather time is money and you don't mind being served by a machine if it is going to give you a comparable or better service. You would rather have that than be waiting around for a machine. You really don't care. And that kind of plays out in our world, in our day to day product every single time because consumers are paying now like weeknights and weekends. Previously your money, like think of it this way, right? mean, something that

 

Adam Parks (21:09)

Time is money.

 

Shantanu (21:34)

this weekend is rather close to the end of the month and for whatever reason if you are unable to get a payment on the 27th which is a Sunday and someone forgets something else happens getting money before the 30th of April is more valuable than getting it on the 2nd of May and machines can collect on Sundays and we've had that experience all day long where know people are paying us on weeknights people are paying us on weekends which is really hard to stop with machines

 

but is extremely convenient for the consumer. So that's kind of the first dimension where we've seen a lot of adoption on the consumer side or change in preference because they can just sell some on their own time. The second is, especially with the rise of like a lot of millennials in the class of borrowers, they would rather be judged by a machine than they would be judged by another human. Millennials.

 

Adam Parks (22:27)

Yeah, that was

 

actually my next question. It's the shame factor is what we call it. And it's a function of do I want to be on the phone feeling ashamed with a live person or do I feel significantly more comfortable having this conversation with a

 

Shantanu (22:31)

It is this.

 

100 % I think and a lot of us again we take We read online so we are not reading from we are not learning as much from other people in classrooms as we are from you know online materials and stuff. So I think there's already a shift where people are like You I don't need you to parent me about this by When I have to pay back her debt, I'm trying to do a good thing the last thing I want is to feel a little bit more guilt and talking to machines

 

in a nice way, in a productive way, but in a slightly more guilt free way goes along with me. So that's kind of the second dimension really, which is like people are coming in and saying, Hey, this is actually very much a preferred mode for me. And then the third is actually the volume rate. think you've every across the industry see like volume of accounts go up, but the value of a single account is actually falling. And as collection industry leaders, everyone is asking like, Hey, what does that do?

 

for my headcount, what does that do for my fixed cost? Do I need to increase it and how does that impact my margins and economics? And especially for like small ticket balances, a lot of BNPLs, a lot of other bills, they are almost cost prohibitive to pretty much any other way. And that's the third wave of adoption that we are seeing where as a consumer who

 

owes maybe only a couple of hundred bucks or not even it is just impossible to serve you well with humans and I think that is where it just works out for everyone to be to kind of discharge a delinquency which is a few hundred dollars over a weekend without being judged so I think that's a holy trifecta that is driving a lot of our adoption

 

Adam Parks (24:31)

It seems like the the consumers are getting more comfortable with it. I think the collection agencies are getting more comfortable with it. I mean, we were we were struggling to deal with text messages and emails like three years ago. Now we're talking about generative AI voice. I mean, think about the exponential change that that requires in the mind frame of not only the agency owners, and let's be frank, right, a lot of agency owners have owned their agency since the 90s. And so they've been at they've been on this ride for 25 to

 

Shantanu (24:37)

For sure.

 

Yeah.

 

Adam Parks (25:00)

30 years and this type of technology has definitely made people nervous early on and I'll admit something to you, Shantanu, because we're friends, like last, 2023 and into early 2024, I had actually circulated a memo within my organization that like using artificial intelligence was a fireable offense. If you're gonna turn in an article, you're gonna turn in content to me that is generated by AI, what do I need you for?

 

Shantanu (25:10)

Yeah.

 

Love.

 

Adam Parks (25:25)

And it was a couple of conversations that I had with you and some of my other friends that are strong in the AI world that made me realize like, I'm either gonna get on board or I'm getting left behind. Like those are my two, I only have two options at this point. And then I spent the next probably three months, I spent about 250 hours playing in prompt engineering, building my own models, building a computer that I could run llama on and actually start trying to really understand it. Cause I said, either I'm gonna,

 

Shantanu (25:25)

Right.

 

Yeah.

 

Thank

 

Nice.

 

Adam Parks (25:53)

figure this out and I'm gonna be at the forefront or some newbie is gonna come in and blow me out of the water because I can't compete from a speed perspective. And it completely changed my entire thought process and methodology as it related to artificial intelligence in general. Now on the flip side, I've got a team of people in the organization that are part of our AI exploration team. And each one of them has a use case that they're focused on.

 

Shantanu (26:11)

Yeah. Yeah.

 

Right.

 

Yeah.

 

Adam Parks (26:20)

And I

 

work with them on a regular basis to help continue to grow their skill set. And hopefully in the next three to six months, we'll get to a point where their skill sets are strong enough and I can have them start training their next layer of people. And we can all start to deploy that. Now, since we started deploying it as of January 1st and using it to generate, I'm going to produce this podcast using AI. I'm going to generate our short videos using AI. This whole platform is going to drive that for us.

 

Shantanu (26:27)

Yeah.

 

Nice.

 

Adam Parks (26:49)

but now we've increased our visibility by 350 % in under three months. It's wild to see the type of growth capabilities that existed and I was sitting here fighting the tide.

 

Shantanu (26:52)

That's amazing.

 

And you're not the only one. mean, there are obviously going to be people who are still going to hold out for reasons that are personal to them. But we are seeing that kind of size, across geography, there are some leaders who are emerging who are actually benefiting from the time. A lot of leaders are coming to me and saying, hey, I see this.

 

increasing the valuation of my business materially because I'm going to kind of be better positioned to acquire other businesses or at least if I'm going to sell my business I'm going to sell it at a far higher valuation. So lot of people, the best leaders are at least like again leveraging the tide to actually overcome what seemed like almost a secular wave of like contracting margins and that's no longer the case and no one needs to be

 

No one needs to feel that way and the best leaders are just like you actually embracing it for their own benefit.

 

Adam Parks (27:58)

Well, it's a scalable situation. In the 2024 debt collection industry report, we saw that we're looking at an increase, 90 % of companies are experiencing an increase in the volume of accounts. At the same time, those same companies are experiencing a decrease in the liquidation of those same accounts, which means that as a debt collection industry in 2025, we've got three options to scale our business. We can hire more people, 88 % are having trouble hiring, 81 % are having trouble retaining the people they are able to hire.

 

Shantanu (28:06)

Yeah.

 

Wow, yeah.

 

Adam Parks (28:25)

We can move into

 

BPO services and we can start to hire in different geographic locations and increase the labor side, the labor pool that we're using within our organization to hire, or we can move into self-service technology. We really need to do all three, but the self-service technology is such a massive push now and the ability for that self-service technology to provide a higher level of customer service in the end. Debt collection is becoming a customer service based business.

 

Shantanu (28:49)

Yeah.

 

Adam Parks (28:53)

and it's becoming an e-commerce business faster than we all want to admit it. I've made the prediction many times now that the debt collection industry will be an e-commerce focused business in the next 10 years. And I think a lot of the technology that you're building right now leads directly into that because having frictionless abilities to collect these payments and reduce that risk level or reduce that shame level for the consumers is going to have an impact in our ability to function as an industry.

 

Shantanu (28:57)

Yeah.

 

Yes, it's true.

 

Yep.

 

Yeah.

 

In fact, I'll take it a step further. It's better than the e-commerce business because it has higher LTVs like your account level lifetime values. They are longer loyalties typically. E-commerce, there is really no loyalty. I if at all you might have loyalty to an Amazon, but pretty much don't have a loyalty to the person manufacturing, even what you buy on Amazon. So there is no loyalty. So think we have industries with like higher LTVs, longer duration relationships.

 

Shantanu (29:48)

And if you're able to partly those advantages and the data that we collect along and get able to partly that into very workable insights that is going to separate out the winners from everyone else. And as a result of it, people who do that are going to thrive really, really well, far more than even the e-commerce players. So you're going to learn from the e-commerce place, but you actually have something that is even better that lets you kind of stay ahead.

 

Adam Parks (30:14)

Well, as everybody's been talking about text messaging and email and like that being the latest and greatest thing since reg F came out, I always go back to it doesn't do anything if you can't get them through a portal, right? If you can't actually collect the payment now, I think generative voice AI provides new layered opportunity for us to actually interact in a frictionless way to collect those payments through a direct methodology, but you can't do that in a text message or an email.

 

Shantanu (30:19)

 

Yeah.

 

Yeah.

 

Adam Parks (30:39)

So continuing to build out the communication channels and the communication libraries that we're using. We've been building out these static libraries and now we've got generative capabilities which kind of changes the dynamic of the landscape for us pretty significantly.

 

Shantanu (30:54)

I completely agree and we do that for a number of large lenders itself, not just collection agencies, lenders. My, my dear friend, colleague, Ragh leads that. And it is so much different. Like previously in order to run a digital program, well, you needed at least a part of like four or five people just focused on digital. You needed someone who would make sure that your emails and text messages are set up, queued and delivered.

 

you needed a different muscle, creative muscle maybe to write out a lot of these things. You need a third person to analyze the success of campaigns and see what is the ROI of it and maybe someone else to marry it on the IT side. In generative, AI has actually made it so much easier to do multiple of these things and it is just seeing such a big lift because you're certainly not staffing in a

 

Shantanu (31:45)

by saying that, this is my voice guy, this is my email guy, this is my text messaging guy, and each one of them operate in silos. You're really saying, what does my Prodigal Intelligence Engine or PIE really talk about this consumer? And then if this is a consumer who prefers email channel, how do I give him a service that is self-service at the same time? Very much consistent with everything else that I'm doing for that consumer. And so the...

 

primary of the first class citizens become the consumers who you're trying to reach. And then the channel, your outreach, your messaging, creatives effectively become something that follows. And you can 100 % embrace AI to do it cost-effectively fast and really well.

 

Adam Parks (32:16)

 

Well, it sounds like you've got your finger on the pulse of what's happening next in the debt collection industry, Shantanu. And I really do appreciate you coming on and sharing your insights today. Any final words for our audience?

 

Shantanu (32:41)

There is so much happening in this industry. think it would be a great first step for everyone to even understand, talk to their friendly neighborhood AI person and understand like in what ways it can be completely transformative for their respective businesses. And you will realize that the opportunities exist where you didn't even dream of it. And the opportunities kind of come in twofold. One is there are new

 

new avenues, new jobs that AI can take on that didn't even exist today, that is going to be an amalgamation of multiple different jobs that you were doing previously, like a lot of the digital stuff that you mentioned as well. So think of it on that axis. And then the second axis is the quality. All this while, and maybe Siri and Alexa spoiled us by giving us really bad answers all this while, people have set like a very low expectation of what AI can do for you. And then the quality bar is like, oh, AI is not going to do that.

 

That is changing. And the quality of outputs that we generate are way higher than what people expect and what they have been trained to expect for the last six, seven years through Alex, Saasiri and the likes. And be open-minded about that realizing that the quality that we are getting is way better than what we've been seeing in the last decade. And as a result of it, it can actually do something that you probably foo-fooed and said it wouldn't be able to do. And that if you're open-minded about it,

 

seeing new jobs and seeing a very high quality bar is been a force multiplier for a lot of our customers.

 

Adam Parks (34:16)

I think those are some really powerful insights and continuing to encourage everybody to explore artificial intelligence as a tool set and what that means in terms of converting those behaviors into actionable intelligence that you can use to collect dollars and at the same time provide a better experience to the consumer. I think that's that.

 

I'm call it the pinnacle of the AI conversation right now around the debt collection industry is really trying to do two things at the same time, right? Collect more money and provide a better experience for the consumers because the creditor that's going to choose collection agency A over collection agency B in a lot of cases now is making a lot of that decision based on how the consumer is going to be treated, how the consumer is going to feel. And I know it sounds counterintuitive. And if I had said this,

 

Shantanu (34:45)

Yeah.

 

Thank

 

Thank

 

Adam Parks (35:04)

Five, 10 years ago, definitely, even five years ago, I would have thought I was crazy. So, I can see what is happening around the space and it feels like the tide has changed and artificial intelligence is something that has, we've hit the point of operationalizing these tool sets and I highly suggest that everybody watch and go spend some time, explore the tool sets.

 

Shantanu (35:10)

Yeah.

 

Thank

 

Yeah.

 

Adam Parks (35:28)

get a feel for where this can add value to your organization, because I have yet to meet a collection agency or debt buyer that can't find value from the use of artificial intelligence. And I'll leave it with one final statement here, which is

 

in the 2024 survey, the other thing that we found is that especially organizations over 100 FTE, but the vast majority of organizations that have made investments into artificial intelligence in the debt collection industry are finding those investments to

 

Shantanu (35:50)

Fred.

 

Adam Parks (35:56)

meet or exceed their expectations from an ROI standpoint. And that I think is really one of those driving points that really will help to get some more people in this industry exploring it and starting to experiment with it and starting to test the waters if you're not ready to jump right in.

 

Shantanu (36:00)

.

 

Yeah, I completely agree and these are the best kind of conversations I have the one with you right now but a lot of other leaders which is what's the potential, what does the future decade look like and I'm always open for it you know no expect I mean it's more about the fun of dreaming what our industry can be a decade from now.

 

that excites me when I jump into these conversations. So if anyone wants to paint a future of the industry, I'm happy. I'm always, always game.

 

Adam Parks (36:48)

Well, I definitely appreciate that for those of you who are watching. If you have additional questions you'd like to Shantanu or myself, you can leave those in the comments on LinkedIn and YouTube and we'll be responding to those. Or if you have additional topics you'd like to see us discuss, you can leave those comments down below as well. And hopefully I can get Shantanu to come back at least one more time to help me continue to create great content for a great industry. But Shantanu, until next time, thank you so much for coming on. I look forward to seeing you out at the CRS show here in the not too distant future.

 

Shantanu (37:12)

Thank

 

Adam Parks (37:15)

And I really do appreciate you

 

coming on and sharing your insights today. You're doing some great work.

 

Shantanu (37:21)

Thank you so much for having me, Adam.

 

Adam Parks (37:23)

Absolutely. And thank you everybody for watching today. We appreciate your time and attention and we'll see you all again soon. Bye.