Building AI Confidence in Debt Collections | Mike Walsh, EXL | Ep. 221

Adam Parks (00:06)

Hello everybody, Adam Parks here with another episode of Receivables Podcast. Today I'm here with my good friend, Mr. Mike Walsh with EXL, here to talk to us about artificial intelligence. How you doing today, Mike? Well, glad to hear it. I really do appreciate you coming back on and having a chat with me today. I know we've had quite a few different conversations through the years about artificial intelligence, but today I want to take a little bit of a

Mike Walsh (00:19)

I'm good sir, I am very good.

Adam Parks (00:34)

different approach and instead of telling people what they should be doing, I think it's time for us to point out some of the pitfalls that tend to come up as you're going through the process of deploying artificial intelligence if you have not really thought through your entire deployment methodology. And where it's such a new technology and so many organizations are now starting to experiment with it, I feel like this is a very timely topic to help companies.

more confidently engage with artificial intelligence and identify those efficiencies and values that can be added to their own organizations. But for anyone who has not been as lucky as me to get to know you through the years, can you tell everyone a little bit about yourself and the kind of the history of your technology experience in the arm industry?

Mike Walsh (01:23)

Well, I got in the industry in 96. Traditional collections was there for most of my career. And I started to see how things were changing. And I jumped, oh, I would say six, seven years back over to VOAPPS direct drop voicemail. I was like, this is exciting. I left there and went to TrueAccord, where I ran sales for a couple of years.

Mike Walsh (01:49)

doing that digital experience. So that's where I learned a lot, right? Like I learned how to take those SOWs and convert them to, you know, those traditional SOWs. You'll make this many call attempts. You know, that's back where we were then. And now, and then how to make them digital. And then I went to Everchain, which was, you know, how to audit after you saw that I saw that need in that space. And I love the solution. And then my first day at

at Everchain, I saw EXL, the demo, and I did not know who they were. They're a huge company, nobody knew. And I saw a bunch of competitors with TrueAccord demos that day. was RMAI already came up to me. But my good friend, Dan Yakumenko, said, let me show you this. And I was like, yeah, dude, I've seen enough. Give me a break.

Adam Parks (02:43)

Yeah,

Mike Walsh (02:46)

And then I saw it and I was like,

whoa. And he's like, do you want to see it in Mandarin? And I was like, what? And boom, he's did the same thing like in Mandarin then. And I was like, wow. that's, you know, about a year and a half later, I jumped over here and it's been great. It's been great. And I've learned so much about AI, what works, what doesn't work. What is the traps? What is how to fix those traps? And

I think it's time if you haven't looked to start looking because I can tell you your competition is looking, you know, the, the, the world is changing. It's changing fast, incredibly fast. So if you haven't, it's time to start. hopefully we'll cover a few things to look out for to make that search for either if you want to build something internally, you know what to look at or, or that search for a vendor like myself, or there's others out there.

a lot easier. and then how to help.

Adam Parks (03:45)

Just so the audience is clear

though, tell us a little bit about EXL and what it is that you're providing to the arm space.

Mike Walsh (03:51)

We provide a lot. we have our payment through solution, which is two way SMS, intelligent email, where it's tracking everything. You can have a conversation with a bot, not really a bot, intelligence. It does voice as well, all under one centralized intelligence, so everything is wrapped up. We provide our clients tons of behavioral data. You can also link your agents to our Smart Agent Assist.

Adam Parks (03:54)

you

Mike Walsh (04:14)

where it's coaching them in real time, it's transcribing them or note taking, it's auditing them 100%, all on the same, on a pop-up platform. It's so easy to integrate with. And then for creditors, we provide auditing features. So you can audit, again, all your agencies without, get every call listened to. And there's gonna be exceptions that need human touch. We're never gonna replace all.

with humanity, for whatever, not going to work. But it's exciting. We have a lot of products and then we're also, we're building things in customer service. building, we do a lot of collection analytics for lots of creditors and debt buyers out there. So we're a global company. It's fun.

Adam Parks (04:58)

Well, you guys are doing some really interesting things. If you haven't seen it yet, I did a five minute pitch with Mike not that long ago going through their payment or tool and talking our way through what that product actually looks like to get those consumers engaged and into your portal to actually negotiate, make payments and and really kind of take that next step.

which makes you the perfect guest for today's conversation because of the breadth of experience that you have as the world or as the world of debt collection has gone digital. I, you know, people will hear me say it all the time. I believe this will be an e-commerce business in the next 10 years and e-commerce will be over the phone. It'll be great. It'll all be electronic commerce. won't just be through the website.

Mike Walsh (05:38)

Yeah, like what is

an e-commerce these days? Like what brick in the world? Yeah. Even your coffee shop, right? Like you can order coffee on your phone. Like, what is an e-commerce? I saw your, what was it? A DCS conference. were like, it is the Amazoning of debt collections. That happened five, six years ago. If you're not there, get there. Right? Like, and now it's the next step.

Adam Parks (05:42)

That's exactly it.

Agreed, and I think that if you look at the ecommerce funnels and you look at the funnels of debt collection, right, where you're dropping somebody in the contact funnels and things, you're going to find them to be very, very similar. And I saw somebody give a presentation just a couple of weeks ago at a private event where they were literally drawing these same parallels. And I do believe that that's what that future looks like. And as we start moving toward that future, you know, a lot of people have been focused on

being able to send out emails, being able to send out text messages, but like Reg F came out a couple years ago now. And so these organizations are still trying to move in that direction, which means there's going to be a big competitive shift for those organizations that have the capabilities of deploying artificial intelligence. What we learned from the TransUnion 2024 report was that collection companies basically have three options to scale their business. They can either hire more people, which they're.

88 % are having trouble hiring, 81 % are having trouble retaining. They can go to BPO services and hire in new locations or you can move into self-service technology. And I think the future lies in all three of those simultaneously. The organizations that control the future are those that can take all three of those channels simultaneously. And through the other conversations that you and I have had as it relates to artificial intelligence, we've talked about some of the things that people need to do and we've kind of organized some of that.

process, but I think people are generally afraid to talk about where it fails because someone who's selling artificial intelligence services to the space doesn't want to talk about the negative portions of what happens when you don't do it correctly, right? Because it instills fear and this is a slow moving industry as it is and you don't want to create any unnecessary fear. But I do think it's important that we address some of the items that...

Mike Walsh (07:36)

No they don't.

Adam Parks (07:52)

people are seeing from a challenge standpoint when they're trying to deploy artificial intelligence into their debt collection operations. And one of the first things that you kind of pointed out for me is the training. Where the training of the model itself and the information that's required to do it. I can tell you from my experience with artificial intelligence, the more...

finite and the more robust of the information that I provide into a model, the better response I'm going to get. And I'm gonna use ChatGPT as an example, because I feel like that's relatable for most of the audience. But if I go to ChatGPT and I say, write me an article about Mike Walsh, I'm gonna get trash, The response that I get is awful, but if I take all of the transcripts and information and articles that you've written and everything that I have about Mike Walsh and I put it into the model and I say, write an article that sounds like Mike Walsh.

I'm gonna get something that sounds a lot better than if I just give it kind of this generic statement. But there's a balance to that too. How much does a model really need in order to understand collections behavior? So from your perspective, what's that balance look like? What's the pitfall of people providing entirely too much information to train the model?

Mike Walsh (09:04)

Right. And, you know, so

I've heard these on sales calls, right? Like, hey, how many months of phone calls do you need me to send over? And I'm like, excuse me? What are you talking about? I don't need any. And they're like, no, really to train. I go, my model's trained. Right. Like, so I would warn people, you know, what is the legal ramification if you're not involved in that debt collection as a direct party?

Mike Walsh (09:31)

I'm

not going to ask you for Mike Walsh's, you know, ABC bank account when I'm trying to work with, you know, a telecom, right? Like, you know, you can't just, so I think that's a red flag. If someone is asking you for your tapes. Now, if it's like a smart agent and you want to get the best quality agents and mimic their behavior, you could anonymize the things, but

But just be careful on how much information you share. Most of the AI products out there do not need a lot of information. The ones that work, the ones that are proven, they've been through that wringer. They've used generalized data. They've worked with clients in the space already, especially the collection space. That's the other thing is I would also be careful someone coming in. Many marketing firms have great products to market, but they're not that collection firms, right?

Be careful, right? It is different. We are highly regulated. We ask those collection questions, right? Like, what does it do? know, are you Mike Walsh, if I have two accounts with the same creditor, are you emailing him twice? Like, I don't know if some of these companies even know that. Like, what are they doing? So I would look for a, the right balance. The right balance to me is, you know, for our collection product,

Adam Parks (10:26)

Agreed.

Mike Walsh (10:50)

I always say it's a dialer file plus an email address. That's all you need. I don't need someone's social security number to call unless you're giving the last four and that's you're required to verify the last four to give access to the account. Fine. We can take that, but we are only taking the last four. Right. Plus I think you, you and I were point you pointed out to me the minimize that that exchange as much as possible today's world, you know, one of the scammers.

Data in transit,

Mike Walsh (11:19)

Yeah.

Adam Parks (11:19)

why have more data in transit than necessary?

Mike Walsh (11:21)

So be careful with the amount of sharing and ask why. Why do I need to give you this? Why do I need this? I don't really want to give you this. Even Bureau data, don't, like some of our clients share it, some do not. You know what it means? Eh, the model's a little slower for a week. Big deal. Right? Like we'll live without it. So if you're not comfortable or if your gut says, why am I giving this to these guys? Ask why. Don't be afraid to ask questions.

Adam Parks (11:47)

Well, and if you're not paying for a product, are the product. So be careful with the data that you're giving out because you may be getting some additional value from. You may be training somebody else's engine with your data and unless you've got some guardrails in place and you're comfortable with it, right? Because a contract is a piece of paper. And how is your data actually being segregated? What kind of clean space is it being held in? Is it truly compartmentalized or is everything commingled?

Mike Walsh (11:55)

You might be training an engine, right? Like, yeah.

Adam Parks (12:17)

And so what's that risk level look like? But even beyond that, my first question whenever somebody comes to pitch me an artificial intelligence product is like, is this actually artificial intelligence? And let's define what that means because I've been pitched a lot of products in the last, let's call it six to 12 months that are if and then statements.

So let's not call it artificial intelligence, you're giving if and then statements on a script. And I understand that, and I think that there was a time and a place for that about five years ago as we were starting to become more comfortable with the usage of artificial intelligence. But how do you address that now? How do you avoid that pitfall of stepping into a product that's not what it claims to be? How do you avoid that?

Mike Walsh (13:02)

You know, this, it's one of the things I've been to a couple of conferences lately and people was like, wow, there's so many competitors here. And I'm like, are there, you know, like, what are they really, what is the AI about that product? Right? Like I think, and look, you know, in my career, I sold it was purely machine learning and I sold it and I use the word AI constantly. And I would announce.

Mike Walsh (13:28)

Hey, never trust the sales guy, right? I'm a sales guy. I would say, Hey, I'm going to use this term and it's not true, but it's just easier for me to say, so I apologize. And that's how I would start my thought. I go, it's really machine learning what we're talking about here in this section, you know? And, and I think AI, if you look up the definition, it's the most bland, ambiguous, nonsensical, anything that can act like a human being. Okay.

So when I'm playing Risk on my phone at an airport or Solitaire, that AI? mean, yeah. I mean, then AI has been around for 60 years, right? So what people are really talking, I think there's two things people are talking about, LLM have open AI as kind of, hey, this is something that I can interact with and it will interact with me.

Adam Parks (14:01)

Is that art if it Yeah.

Mike Walsh (14:18)

Agenic AI or people are dropping it to Generative AI or Agent AI. To me, these are things that are making decisions based on the inputs so that humans don't have to get involved and do it. Now, they're in debt collection that is good and is bad. And there's a balance there too, right? We are trying truly Agenic AI

Mike Walsh (14:45)

in the communication space, but we're more in customer service right now. We've, we've, we've launched it because debt collection is so regulated. So to get a Agenic AI to follow regulations is it's a challenge and we've been working on it for years and we're very close. But I would also say that's not necessarily probably what half the things in the market are. but it's still.

There's other ways to do it. And there's plenty of ways to improve your collections. Just be careful of the buzzwords, right? Like everybody's using the buzzwords. You can find a definition of any word, any AI term that will say, this is it. Trust us. It's going to be great. I would ask, you know, is this all yours? Are you piecing together other people's products? Right? Like that's a great question to ask, right?

Adam Parks (15:24)

That becomes the next pitfall. It's the who actually is running this. So is this chatbot that I'm engaging with just an old version of ChatGPT with a new mask on it.

Mike Walsh (15:51)

Right, any of the other, you know, there's hundreds of LLMs out there and many of them are very good. And, you know, we started with one and then we built our own. And then we built, we have a patent for one in the insurance business. We have our own collections, LLM. Because...

We want it private. We want to keep that information here. We don't want it going on the internet. We don't want access to a third party having it. That's sensitive data. Plus, control, tweaking, updates, it's all on us. And our solution is communicating with itself, not 10 other solutions. So if one of them fails, what do you have? Incomplete. What happens then? What if we built our own compliance engine?

Built all these different engines within it. So you just want to see what that tech stack is. then as AI grows, and this is something I don't think we talked about before this call or this podcast is computing power is becoming a commodity and it's going to get more and more expensive is so we've bought tons and tons and tons of computing, computing power. Like, so we're

Mike Walsh (17:02)

we're moving our solution to Nvidia for speed and computing power. But we create an instance for each client. I would ask, is my data commingled in any way? Now there's anonymized data that it learns from $100 accounts, gonna look at other $100 accounts, but it doesn't know where it's from and doesn't know who it's from. That's anonymized data. But just be careful there too.

I would want to know how my PII is segregated from other PII. What products, exterior products access? Because your clients are going to ask. If you're a collector, collection agency, they're going to ask, wait a minute, who owns this? What is this product? Let me see the model governance. If you're using a vendor, ask for that. There should be at least a responsible AI policy.

that everyone should have that explains, we're not using this collection data to sell loans. that's, you know, you got to be careful and make sure you understand what it's the use of that data is. And then models, you know, I think you and I were talking the other day, back in the day, zip code analysis was a big part of collections, right? Like zip code analysis today will get you sued.

It will get you sued. And I've heard about the stories of people being sued, right? And I'm sure you have too.

Adam Parks (18:25)

Yeah.

Mike Walsh (18:26)

So, yeah.

Adam Parks (18:27)

It's a very different animal, right?

In 2006, Credit Max had the debt sales system where they were selling accounts on a zip code basis. Today, I don't think that you could segregate accounts in that way without really falling down the redlining path.

Mike Walsh (18:40)

getting fair lending. Yeah.

Fair lending, fair, like you couldn't settle by zip code at different rates, you know, and I've run into people like building model that like, yeah, yeah, I was going to break down my zip code. go, I get rid of it. If I were you, I'm like, I would not go down there. So those are the questions though, you have to ask. And, and I don't want to scare everybody because this process isn't that hard, but

Mike Walsh (19:07)

Hey, you wanna know where your data is going? You wanna know is the product owned by this person or is it pieced together? You wanna know where it's stored, who's working it with? I love when I run into people and they're like, yeah, all our developments in South America or Philippines or Australia or Ireland has a ton or India. But we never send data outside the US. Wait a minute.

How many developers do have in the US? we don't have any. When it breaks, how do you see if Mike Wallace's account and Adam Parks's account somehow became combined? Who looks at it? Because they can't. You can't send that data across. So I think that's another important thing is make sure. mean, look, everybody uses offshore, nearshore, onshore. There should be a mix of those things. You should have people in the US.

Adam Parks (19:33)

Who? Yeah, who's troubleshooting?

Mike Walsh (19:59)

And for cost benefit, you should have people across the globe. And that's what you want, too. But make sure there's at least one or two developers in the US on a product before you buy it. I would say, where is their location and where is my data store?

Adam Parks (20:12)

I think that's a pretty good piece of advice for organizations that are trying to leverage this because the location of data, I mean, it wasn't even five years ago where the push to the cloud really became critical mass. And a lot of organizations now, if somebody tells me that for years it was like, you put your data in the cloud and people are almost looked down at, or organizations were looked down upon because they were using the cloud. And now if somebody tells me that they're managing on-prem servers, I'm like, really, why?

Mike Walsh (20:26)

Right.

How'd that work?

Adam Parks (20:40)

Are your, is your internal

IT staff like better suited than Amazon, Google or Microsoft to protect and manage that data at this point? I mean, I'm in Florida, so it really has not been an option for me because with the power flickers and everything else, like trying to host data in Florida is a waste of time. But now, but think about that shift over time.

Mike Walsh (21:00)

You know, with the three-clock green spoon, right?

Yes, you're right.

Adam Parks (21:03)

Well,

what's the next five years look like, Mike? Like what happens over the next five years? Are we going to start looking down on those organizations right now or early on? People were looking down at conversational AI as compared to other use cases of artificial intelligence in the space. But what happens over the next five years? Does that completely flip on its head and you're going to start looking at organizations differently because they did not deploy artificial intelligence early enough in the cycle?

Mike Walsh (21:28)

You know, that's my question for you, right? Like, Adam, like with the amount of debt, both like medical credit card, you name it. And now, now we see the ugly mortgages going up, which is scary, right? Like, so you have this massive amount of debt and then you have, you have, it's more and more competitive to get people employed, right? In a call center.

Mike Walsh (21:50)

You lose people for a quarter an hour and a half a tank of gas like or a third of a tank of gas because gas is so expensive really So just maintaining your your your people is so hard even the ones you have that are trained let alone hiring new ones I mean you have to interview 10 to get three then you have to go send those three through the class and hope you get one Yeah, hope you get one out of it or is it 10 100 to get 10 to get one like depending on where you are it's hard for I like

Adam Parks (22:09)

Hope they show up at all.

Mike Walsh (22:19)

I think you're right with your first statement. You have to find balance in this space now. You need that digital outreach. The virtual agents are coming. They're coming fast. And people think it's like five years from now. I'm telling you it's this year. They are coming that fast. Maybe, maybe, yeah.

Adam Parks (22:39)

Well, think about the speed of things.

Look how fast the internet came into play and ecommerce came into play. That was let's call it a 10 year cycle before it became somewhat ubiquitous. And now if we look at the speed in which the advancements in technology are happening on the models, it's wildly fast. And like now we're looking at this like exponential curve, we're looking at massive amounts of data being added. I was doing an AI Hub podcast with with Tim Collins. And we were talking about just

how much data has been created since 2007 in the invention of the iPhone, right? Like in the last 20 years, right, or the last 18 years, has been exponential in terms of the volume of data that's being created, even on a daily basis. And the only way for us to actually understand that volume of data is to start to deploy this type of technology and tool set. And it's becoming ubiquitous at a much faster rate in its...

Mike Walsh (23:15)

Yeah.

Adam Parks (23:39)

going almost straight to the hands of consumers like these types of tools. If you look at PCs, for example, computers, that was a long curve, but it started really B2B and then they started making it available to consumers. If you look at IBM, international business systems, not international like have a computer day. It's a very different animal. And we think about what that, it started with the business and it moved up, but artificial intelligence has flipped that on its head. And now the consumers had it in their hands faster than the businesses this time around.

And now those people that have started using it and playing with it are bringing it into their businesses and starting to use it.

Mike Walsh (24:14)

Right, and we always think of it in collections as the collection part, but it's other, we don't program it. Like Excel stopped hiring programmers, AI programs for us. We tell it what to do, we watch it, we write it, but there's so many back office aspects of what's gonna change. It's not just consumer facing that's gonna change. Audits are gonna be automated. There's no point in sending a person to checkbox it.

Mike Walsh (24:40)

Right? Like where machine could do every call you have in your call center in minutes. Right? Like it could live in the case. Yeah.

Adam Parks (24:46)

And then just kick out an exception report for a manual person

to review. These are the things that we didn't understand or felt that they were close enough to the line.

Mike Walsh (24:50)

Correct. And it can give you,

and it can measure like their sentiment of the call. Hey, they were angry, then they settled down, then they got angry again. Right? Like it can measure and give you a confidence score. Like that's absurd. Like when we, when I started in this business, that was a dream, right? Like we didn't have that many QA and then QA became important. Now it's, it's going to be automated. It is. so another thing to do when you're

Adam Parks (25:16)

And layered,

you've got the you've got the voice, we're talking about the voice compliance. And now you've got groups like CollaborationRoom.ai that are layering on top of that with video recognition. And looking at the not only the not even necessarily the emotion of the consumer, but the emotion of the collector themselves.

Mike Walsh (25:18)

Yes.

Rep,

yeah. like the data privacy of that area they're working at, it's amazing. Like I talked to those guys at ARMTech and RMAI, I was like, wow, so cool. So it's getting better. I think there's a lot of people resistant, and I would say that's something don't do. Don't be resistant. It's time to learn what's out there.

I used to say when I worked at TrueAccord, I'm the least tech guy working in a tech company. I made fun of myself and all the reps there would make, here comes the old man trying to, but it's not hard to learn. There are plenty of things. Go to LinkedIn. They have some great tools to learn about AI, generative AI. They have courses that are free. It's a little bit of a commercial for Microsoft, but still valuable.

What you start thinking about is, okay, what's the problem at my agency? What do I want to solve? Right. And you start breaking it. And I think one of the things I want to tell people is don't jump all in all at once. Because if you pick the wrong vendor, yeah.

Adam Parks (26:38)

It's another one of the pitfalls.

Is just cannonballing your way into the pool without thinking your way through it or feeling like you got to do consumer first.

Mike Walsh (26:43)

Don't have a sales guy. Yeah.

Like don't let me come in your office and sell you every tool I have at once. And yeah, I'll make great commission, but you'll hate me, right? because it's too much too fast, right? Like I would say, and the way I do it is I say, here's what we got. What do you need? What's the top priority? Prioritize them. You don't know? Take some time.

Mike Walsh (27:09)

look at the numbers, look at the data you can get, right? Like, and say, okay, this really is our top problem. You know, maybe it's our call, our average call time's gone up since there's so much debt, whatever, I don't know why, maybe you need an agent assist more than you need a digital, right? For now. Or to reduce those calls, you want to go digital right away. But I would say, you know, we generally do our implementations in phases and then you get ROI in each phase.

first phase pays for the next one. And another thing is don't write a big check. If you're a collection agency, you should not be writing some AI company a big check. I know my product works, right? Like I know it's gonna work. I am 100 % confident. I don't need you to write me a big check. You're gonna pay me over time. We're gonna be partners for 10 years, 20 years. I don't even know. Yeah, because...

Adam Parks (27:59)

Let me ask you this, when we talk

about the confidence level of those people that are starting to get involved, do you think it's worthwhile before they start experimenting and concentrating on this consumer facing type of experience that maybe they just go sit down with chat, GPT or cloud or any of the ones that really any of these AI models and just start developing some level of confidence or comfortableness with the tool set itself?

Mike Walsh (28:28)

I mean, think right, like I don't know if I trust my research, you know, on ChatGPT right? Like, you know, but I think it's interesting to, it's like, it's interesting to say, you know, chat, pull up ChatGPT and say, give me a collection call script. And it will give you a script, right? Probably something somewhat similar to what you use, not as refined. It'll be generalized and say, give me a collection dispute call script.

Adam Parks (28:28)

I realize that they're two different a baby step somewhere.

Don't. Don't.

Mike Walsh (28:56)

And that's going to be a little bit like you said, you know, play with it. And that's when you can kind of start seeing. And we see this in our implementations. Like it takes a little bit, like two months, right? Like by week six, people like, let's go, let's go, let's go, let's get this thing running. But then they go through testing and they see the detail in which you have to like break these things down to steps. they're like, thanks. Thanks for doing the test. Now we get it right. Now we get why we got here. So that's another part of the phase approaches.

you know, test, I would play with, you I did, when in ChatGPT, Daniel Green over at Everchain, he kinda like, I'm like, I don't need that stuff, you know, I've been doing this forever, right? I'm a dinosaur. He's like, no, no, man, play with it, like, try it. And I played with it and some of other guys I worked with, you know, cutting down their emails and I'm like, oh yeah, this is kinda slick. And there was an add on to Microsoft.

I don't know or whatever it was. It was wonderful for emails cutting them down to like one sentence. Even before Copilot, but I'm sure that's what Copilot is. probably follow them and that's what it is, but I'm not really sure. But stuff like that, was like, I can see how much time this is saving me. And time is effort, effort is money. That's what you're trying to do.

Adam Parks (29:58)

Yeah, copilot and you get Gemini for Google. I mean, there's Okay.

You

Mike Walsh (30:17)

Start with the outcomes you're trying to fix and work your way back. And don't, I think people also get scared of the data. Like, our system, even if you have an old system, like that is not on a cloud and is, you're a small agency. Would you rather send two emails or two letters a month?

and hope they get delivered or would you rather get an AI program to work that account, you know, using data, behavioral data, give you all that data back so you can make phone calls based on what's actually happened with the consumer that you're tracking everything. I mean, that's, that's, that's really what you're doing. Like two letters first AI that works 365, you know, works on Christmas at midnight. You know, like if someone wanted to pay the bill, cause they got a bonus.

They could. And then.

Adam Parks (31:08)

consumers

are getting more comfortable with the tool set, right? So if from a customer service standpoint, you are looking at these consumers that are starting to get more engaged with these and they're starting to get more comfortable using them and it's only a matter of time before that customer service confidence starts to flow into the debt collection space and the consumers actually start looking for this similar to the way that they've been looking for other self-service options.

Mike Walsh (31:10)

They love you.

And.

What would you rather do? Talk to a stranger about your debt, you know, your financial hardship over the phone. You can't see them. You don't know who they are. Never heard of them. Or go to website, Google them, make sure they're legit. Go self-negotiate. Pay something you for. Not get talked into a $125 payment because they need to make a bonus. Just pay the 75 you can afford. And if you can't afford it, turns out you can pay 65 to renegotiate.

whenever it's convenient. You don't have to do it between 8 a.m. and 9 p.m. You can do it at 10 30 when your kids are asleep. Like when you get to breathe. You know, actually your kid probably isn't asleep at 10 30. But like that's what I go back to that panel you were on. You know, that's what you were talking about, right? you, I gave some, I was talking to an agency the other day and I said,

Adam Parks (32:12)

I understand. I understand.

Mike Walsh (32:33)

you know, what you're going to turn into your client. And this is probably a big thing is don't do this without talking to your clients and seeing if they're interested in the answers they are right. They want you to do these. And, and basically I was saying, you know,

Where do you think this is going to go? We need to be customer centric. So no one's going to send an RFP anymore that says, is your primary strategy, your secondary, and your terche. Because the consumer in each level is going to determine their own treatment strategy. That's behavioral.

Adam Parks (33:05)

Yeah.

Mike Walsh (33:07)

the customer is going to determine the.

Adam Parks (33:08)

That's it. That's

an interesting approach to it is that it is so different now in the way in which the behavior of the consumer can drive that work treatment at the various levels of work.

Mike Walsh (33:21)

And it's hard to think that way, like from a guy who started in this industry in 1996 on the phone was first two months of my, had to make phone calls and do it.

You get payments, you get excited, you think this is it. What are the prime times to call? Blah, blah, blah. With a cell phone, is there any prime time to call? I don't know. I answer my phone totally randomly based on if I feel like, oh, this could be funny. And like this morning I saw a number I didn't know who it was. It was a prospect going, hey, I think I have a phishing attack. Can you check if you send me this email? And it was my quarantine nonsense. But, but.

Opposed to like that was just totally random because it didn't have her number with her name so I think it's We're in a different world and it's time for different ideas and and again There's a way to do this. It's clean like Listen to this podcast ask questions Talk to people in the industry Industry expertise and industry knowledge to me

There's a lot of startups that come jumping in. And I think they come from marketing or they come from other worlds and they know everybody sees the debt numbers and they're trying to jump in. But man, make sure they understand.

Adam Parks (34:33)

But if there's no underlying experience of the industry itself, there's some red flags that need to be addressed. If you don't have an underlying understanding of what debt collection is as an industry. And I think that's the kind of the final message for us today is like, you got to build that comfort and confidence in yourself, like play with some tools, actually go through the process yourself.

and then go through it. And if you're being asked by a vendor to do something that makes you uncomfortable, that's a red flag. If you're not comfortable doing it, like that's a red flag. Ask questions. Ask questions, right?

Mike Walsh (35:07)

Yeah. Ask why. Why? What are you going to use it for? Right? Like, yeah. And we can ask that like,

look, I get asked it all the time. Like, wait a minute, what about this? What about this? And we were happy to answer those questions. We are happy to show any word in our scripting that is a dispute or consider dispute. I'm I share that with my clients. I share everything. Like there shouldn't be some secret AI sauce that

Look, we have people that will be telling me, you want to build it? Go ahead, copy it. By the time you build it, we're going to have something better anyway. That's how we look at it.

Adam Parks (35:38)

Yeah,

yeah, good luck trying to build I mean, look, it's not easy to build it yourself when it comes to artificial intelligence. I've played that game internally before I wanted to put anything out to the web or gain any level of confidence with third party models. It's a lift. It's a big lift. And we saw in the survey, or in the debt collection industry report this year that there are a lot of companies that are over, let's say 100,000 accounts, like larger organizations that are

Mike Walsh (35:44)

Right.

Adam Parks (36:08)

actively trying to do that. But it seems like the majority of organizations really are looking at pre existing third party solutions and trying to find the right fit for their needs. And if your confidence level isn't generative AI conversation, and it's looking at some back office opportunities, like that's a real opportunity as well for you to dip your toe in, actually feel the impact that can be felt on this and then start to think your way.

Mike Walsh (36:29)

I agree.

Adam Parks (36:37)

through it. any final advice for our audience today,

Mike Walsh (36:41)

That's a good one. I would say just, know, the great thing about the debt collection space is there's plenty of people out there that are willing to talk, right? Like have those conversations, know, attend the conferences. If you can invest in a conference, if you're going to go down this route and have those conversations, you know, it will, I mean, I will say don't be afraid is my number.

Adam Parks (36:51)

statement.

Mike Walsh (37:04)

It is time to do it and you can do it. You can have a great success. It's not going to get you. If you do it right, you're going to have a better, more efficient operation that is more profitable and customers like better. And to me and clients, right? If the clients customers like it better and you're more efficient, you're going to be producing better. What's wrong? So it's time.

And you just look at the players in the space and evaluate who's the best fit for you.

Adam Parks (37:37)

future is now that's what I hear the future is right now. I agree. Mike, I really do appreciate you coming on and sharing your insights with us today. I know with all of the experience you have getting to work with so many different companies, I appreciate you continuing to come back and help me create great content for this industry about artificial intelligence.

Mike Walsh (37:38)

It is, Mr. Anderson. The future is our time.

Adam Parks (38:01)

For those of you that are watching, you have additional questions for Mike 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 cover, you can leave those in the comments below as well. And I'll bet you I can get Mike back here at least one more time to help me continue to create great content for a great industry. But until next time, Mike, thank you so much. I really do appreciate you coming on and sharing your insights.

Mike Walsh (38:14)

Sure.

Thank you, sir. Get some sleep. Congratulations

again on the new one.

Adam Parks (38:24)

Thank you so much, I greatly appreciate it. And for all of you that are watching, thank you so much for your time and attention today. We'll see you all again soon. Bye everybody.