Adam Parks (00:07)
Hello everybody, Adam Parks here with another episode of Receivables podcast today, very lucky to have with me Pete Klipa, the Chief Client Experience Officer with Harvest Strategy Group to talk to me about well kind of the current wave of what’s happening across the debt collection industry in terms of the use of artificial intelligence and other technology tools and how that is ultimately impacting a change in our dynamic.
It used to be we have a problem and we would could throw bodies at it, get more people on the phone. But that’s not really the way of the future. And it’s sure not the way things are going right now. So Pete, thank you so much for coming on with me today. I really appreciate you sharing your insights.
Pete Klipa (00:55)
My pleasure. Looking forward to chatting about this topic. It has many angles to consider.
Adam Parks (01:01)
So Pete, for anyone who has not been as lucky as me to get to know you through the years, could you tell everyone a little about yourself and how you got to the seat that you’re in today?
Pete Klipa (01:09)
Absolutely. I’ve been in the collections and recovery space for the better part of 25 years. Several different verticals during that time, including public utility, credit card, had a stint in credit counseling where I got an opportunity to work with a lot of the banks in that nonprofit sector. Currently I’m with Harvest Strategy Group.
I’m the Chief Client Experience Officer and I’m in my fourth year at Harvest. Harvest is an outfit that does work for regional creditor clients as well as debt buying clients and we have a master servicing role and we’re like a lot of folks in the industry really seeing some outstanding growth in the current environment.
Adam Parks (01:52)
Which is part of the reason I think you’ve got such a unique perspective, not just your history in the space, but kind of your current role in helping organizations to manage those vendor networks and to be part of that entire process. So I feel like it gives you really good visibility upstream and downstream as to how these things are actually happening.
And we’ve seen this constant increase for the last few years in our survey results that it’s harder to hire people and it’s harder to retain the people that we’re able to hire and with those challenges in place or those challenges still be being very real It’s difficult for us to throw bodies at the problem and as you’ve seen the an increase in the volume of accounts that you’re managing, what have you seen Pete, organizations doing to offset that need to hire?
Pete Klipa (02:44)
Yeah, absolutely. It’s a dynamic where, you know, I think there’s an expectation, you know, from clients that we find efficiencies and we have to keep pace, you know, with those efficiencies to try to continue to keep the lid somewhat on, you know, the extensive hiring because the hiring process, the training process, it’s expensive.
You know, there’s still the dynamic, you know, in a post-COVID world of, you know, the whole hybrid environment of working from home versus working from an office. So I think as all of us in the industry navigate that, being able to find efficiencies to not have to do as much hiring in the environment and the growth that we’re in is a target. just like a lot of things that you target, there needs to be some parameters and some rules around it.
And particularly with something like AI, it has outstanding potential application and even some really market actual application. But at the same time, there’s some rules to consider around it and it can’t be unlimited, it needs to have human involvement as well as client and regulatory sign off and consideration.
Adam Parks (04:04)
We hear a lot of conversations around the ideas of deploying artificial intelligence in the space. And as I’ve gone down the research rabbit hole, looking at a variety of what’s called academic reports and things, it sounds like some use cases are having deeper impacts than others when it comes to the deployment of this technology, whether it be for an agency or for a servicer.
Are there any specific use cases that you’ve seen that have been more impactful than others in these early stages of AI deployments?
Pete Klipa (04:36)
Absolutely. I think a general comment about it is that to the degree that it can replace the use of manual tasks, that really does have a sweet spot kind of element to it because manual repetitive tasks are things that can be displaced with something that’s just a lot more efficient. it could be in the summary of data packages so that you get a quicker feel for what a large body of data is telling you or even a large body of data that’s already summarized, what are some of the AI findings around that. So that can be very useful.
I have a couple of examples of use case that I could mention that are, you know, things that happened with us at Harvest, you know, and one of them is kind of a personal thing with me, you know, I don’t know if you guys remember the old legal black books, you know, for non-attorney laypeople. There used to be in the industry a little black book that was like a reference book around legal terminology, you know, mostly things that attorneys, you know, would have pretty close at hand, but, you know, when they start spotting awesome things, you know, around Latinisms and, you know, courtroom detail that sometimes is beyond the untrained eye.
I used to use that, that black book, you know, for reference, especially when I was new and early in the industry. But recently, you know, kind of got turned on to a program called Perplexity that has the ability to reach into, you know, website data that has detail, you know, at the various court levels within states and jurisdictions. It’s a very handy reference tool to understand local happenings in legal.
And legals like politics, it kind of plays out at the local level. And it’s important to have a base of understanding based on what’s public information out there that a program like Perpexity can pull upon. So that was one application had some data that was already summarized from account level data on our side at Harvest that needed to be, I guess for lack of a better phrase, had to be packaged up in such a way that we were pulling information from desperate spreadsheets into one spreadsheet that had to be consolidated for a particular client. Some people at Harvest were able to write, you know, very quickly a program to pull a large amount of spreadsheet data together into a package.
And, you know, probably the human time to pull that together would probably have been two or three hours, and it was reduced to two or three minutes, you know. So that type of efficiency, I guess, is an example of AI in action, you know, to deliver a result, that resulted in a lot less human time. So those are the types of things I think we as an industry are trying to find those nuggets, those opportunities, because the more you can do those, the more it can be duplicated. And through that experimentation, there’s an opportunity for expansion.
Adam Parks (07:56)
What I’m hearing is that it’s not just about the debt collection specific use cases. It’s about those general business use cases that transcend industries and really, you know, kind of individual businesses, because what I heard there was about moving data, right, not necessarily just from a collection agency standpoint, making litigation decisions, for example, but going and looking at our businesses as businesses and not just debt collection businesses and identifying some of those opportunities where we can really deploy some tools to improve.
I mean, Perplexity, I’m in it every day. I love perplexity from a deep research standpoint. That Sonar model is one of my favorites that in the tools availability to use additional models like ChatGPT or some of the Claude models I think makes it a really versatile tool.
The level of detail and description that I’m able to retrieve from the web or from research documentation, et cetera, has been really impactful, at least for me, from a research and an understanding standpoint. So I definitely get where you’re going there. As you’ve seen organizations start to deploy AI, let’s say vendors, whether it be law firms or agencies,
Have you are you mostly looking at the use cases from a debt collection standpoint? Or are you looking at their use of artificial intelligence across the board kind of the way that you’re viewing it in your company?
Pete Klipa (09:23)
It’s a little bit of both. There’s a very high importance for client review and sign off on AI use cases. For sure at the vendor level that we’ve managed and we as ourselves a vendor and a provider that sort of serves as that intermediary with vendors to clients. There’s a lot of client interest in it, but there’s also this need, you know, to make sure that there’s protections, you know, particularly around account level data.
So I think in a lot of ways it’s leading toward, you know, more summaries of existing data that’s already summarized, you know. So in other words, it’s not involving PII, you know, but it’s maybe AI use of, you know, review of summarized data, I guess, which, you know, puts it in a little bit of a safer place with the deliverable that comes from that.
There’s client interactions in a way of like presentation material that is lending itself to a review, the AI with analysis that ultimately can get packaged for a client review that is faster than a human putting that together initially and maybe 80 to 90 % of it is already in kind of a near final form. And then you move forward with it, you know, with some tinkering and give it into the final form with the last bit of human touch.
You know, there’s applications in contract review. You know, when you think about, you know, managing contract detail and, you know, the type of review that that involves with a lot of reading. It can assist with interpretation around that with the subtleties and differences in a contract. Even policy review or policy rewriting or creation. It has the ability to synthesize some things from that and get it into a form that could then be reviewed without a ton of heavy lifting and then more of a fine tuning in the end state.
You know, there’s lots of things around, you know, the language model, both written and verbal. You can cast a wider net around that, you know, to understand what is being said and reviewed here and where do I need to focus my attention from an exception standpoint. So it’s a streamlining while casting a wider net at the same time so that you can focus on the things that are most important in the review of that compliance and quality activity.
You know, the applications are endless and, you know, with that becomes a lot of potential, but also, you know, the need to make sure that we go about it the right way with, you know, foundational controls and then building from there to the next level as you kind of get it to the place where there’s a comfort level with which you complete and then it lends itself to expansion.
Adam Parks (12:25)
Are you seeing that the organizations that are best performing with these tool sets have kind of reworked their workflows versus bolting on tools to their existing processes? And where you go so deep in your due diligence and your auditing processes, has that been your experience for high performers?
Pete Klipa (12:47)
Yeah, you the integration with, your, your enterprise system is really important. You know, it’s a lot like digital that way in the sense that, you know, I think digital has shown that it’s at its best when it’s integrated with my system, you know, and I think that’s, that’s true of AI as well, because, you know, it has to be established in such a way that it’s, it’s not on the outside of the system.
It’s in a way kind of tied into what the system is telling you, you know, and that could, you know, help in reports and analysis of those reports of what those outcomes are so that you can, you know, kind of focus on review and decision making, you know, from what AI is kind of setting the table for. And there’s some efficiency to be gained with that for sure as it gets more and more sophisticated.
Adam Parks (13:41)
Do you think that we’re focusing too much on the artificial and not enough on the intelligence portion of it? I hear so much conversation around, we’re gonna be able to call cheaper, we’re gonna be able to do this less expensively, but it feels like we’re kind of missing the boat there, when the intelligence portion, the decisioning engines, the machine learning at scale, those kinds of things seem to be where we can find the largest impact in the short term. What has your experience been?
Pete Klipa (14:13)
Yeah, you know, I may be a little bit old school on this and I am Ohio based. So you know, there’s a little bit of a high state Buckeye in me as it relates to this. But, you know, a famous high state coach from way back, Woody Hayes, you know, wrote a book, you you win with people.
And, you know, I do think collections, you know, for all its system advantages that continue to come into the industry, structure changes that we implement, I do think at the end of the day, it is a business and an industry that you win with people on. And that’ll be true, I believe, even in an AI environment, that human intelligence, sort of side by side or parallel with artificial intelligence, I think that’s where our industry examples will be at their best and are at their best, where you have that regulating human function that comes in and says, is this really what we want to do?
Does this pass the test of my history, my knowledge of the industry? Because sometimes it will take you down a rabbit hole, and you have to be careful about that. It can be a time stomper if you’re not involving some human elements to it. I did hear of an example where, you know, from a programmatic standpoint, a collections partner kind of turned it over to AI, you know, with a database of some sort.
And, you know, the database was somehow erased and then it went so far as erasing the backup database as well. And, you know, that’s kind of a double whammy and certainly would be, you know, resulting in a loss of productivity with that outcome. So, those are types of things you want to avoid. And I think with experimentation comes documentation and kind of understanding and sort of sharing that understanding from a training standpoint so that others can benefit from something that’s so new for many of us.
Adam Parks (16:19)
I always encourage people to write an eighth grade experiment document along with their test. Look, just defining your hypothesis and where you’re trying to go with this thing and what you’re going to experiment for. And I also find that defining success before we begin the experiment helps us to kind of set our bearings as to whether or not this met our expectations because so often our expectations change on the fly as we’re going through this learning process.
And it sounds like you kind of gave me one nightmare scenario there with an organization deleting the database and the backups. But as we think about agencies and law firms and other collection vendors on a higher level, where do you think they’re failing or missing the opportunity from an artificial intelligence deployment standpoint? Is there something that you feel like they’re maybe avoiding that could have serious impact?
Pete Klipa (17:15)
Well, it’s almost like there’s a tipping point of this where, you if you’re really going to set up an environment that, you know, is safe in terms of PII and, you know, being able to protect, you know, key client information, that sort of thing, you’ll almost have to make that investment to get things set up on your side of the firewall, if you will, so that you don’t have to go outside of it, but you still can benefit from the tools and the technology that some are benefiting in the industry.
And that setup, it’s not cheap to get that kind of expertise, but I do think that it makes sense to make that investment at the businesses at a certain scale, because the environment we’re dealing with right now, it reminds me a lot of the environment that we had, you know, really pre CFPB, you know, when we had so many accounts that were coming, you know, and just not enough time and not enough hiring ability to grab them all and give them the necessary attention that they need, you know, so AI can help a lot with that.
Particularly, you know, with things like scoring, you know, it can lend itself to some input around the scoring so that we’re working as smart as we can in light of all the accounts that are flowing through the system right now. I don’t know what that scale point is, because we have small vendors in our space for sure, and we have large ones, but I don’t know that anybody can be exempt from it, because the industry is changing and you sort of have to keep pace with that, with a commitment as a company to, to gather some knowledge of it so that you can apply it in your world.
Adam Parks (19:06)
Orchestration seems to be one of those key pieces that we’re finding value in and I know that there’s more than a few use cases where we can find value but it felt before like people talked about omni channel but without that coordination engine without being able to collaborate across those channels it felt more like disparate channels with you know similar context
What have you seen in terms of organizations results as they’ve coordinated the efforts across their various communication channels?
Pete Klipa (19:42)
You know, I think with digital, you know, we’ve, we’ve seen a whole lot of difference that it made, you know, in the out of the gate, you know, as, the digital adoption experience, you know, it was clear the folks that were using digital and the ones that had not yet adopted it, you know, so it became a differentiating factor with AI. I don’t know that it’s as far along to be able to evaluate how much of a difference it can make in results.
You know, we’re having more & more conversations, you know, with our vendors about it because, we expect our vendors just like we expect of ourselves to have a process and a program of innovation and AI is part of that, you know, and if we ask them, you know, what are the things that you’re pursuing in the way of improvement initiatives that we as a company can benefit from and our clients can benefit from?
And they kind of just have a blank look, a blank stare. You know, that’s not the answer that we want. So, you know, there has to be this thought of what am I going to commit to innovation? How am I going to get my business to the future so that when we look at it five years from now, it’s going to look different, you know?
So those types of things I think are part of the dialogue, but, you know, it’s the type of thing where I think the AI is still pretty new to say, hey, this is paying dividends here because you also have this balancing effect with the clients and their expectation for sign-off and review that you can’t bypass that. That part is in a way a type of regulating mechanism.
Adam Parks (21:22)
Interesting and as you’re looking at organizations that are providing services and doing it with any type of AI in their workflows and now we see that 93 % of debt collection organizations are deploying some level some type or exploring artificial intelligence. How do you distinguish between an AI enabled partner and an AI dependent partner?
It feels like we’re going to start and we’ve all read the articles about you know, ChatGPT and brain usage and all of that. But how do you how do you watch for that going into the future?
Pete Klipa (21:57)
Yeah, that’s a good one. I’m not as worried about the dependent piece. I’m more worried about the flip side of that, which is a lack of dependence or an independence from dependence. We think that there are practical uses, right? And as those practical uses get more documented and increasingly safer with the, them on their way to perfection. You know, I think it’s a conversation, you know, around what are you doing with this?
And we’ve seen, you know, differentiation, you know, amongst vendors that we’ve talked to in terms of what they, what they’re doing, what they’re planning doing, and some, you know, that are a little bit slow out of the starting blocks. And, you know, I think what we try to do is understand better, you know, from a best practice standpoint, what some of the vendors are out about doing, you know, asking questions like, Is it safe? Is it approved? How do you know that?
That type of thing versus some of these others that have said, hey, I’m just not willing to commit to it because I see it as too risky. That may not be the right answer either. I think the best place is someplace in between.
Adam Parks (23:06)
Have you started looking at different KPIs or measuring vendors or your own performance in a different way as you’ve started deploying this technology and watching it be deployed throughout your network?
Pete Klipa (23:25)
Yeah, I mean, one of the best KPIs to look at kind of at the highest level is, you know, the financial liability of the vendors. You know, if they are, you know, working through this the right way and gaining some efficiencies, you know, there should be evidence of cost savings. You know, there should be evidence of holding the line on hiring growth with expenses while, you know, achieving some top line increase in revenue, that’s probably the ultimate KPI measure, that profitability as well as solvency that you see in a financial statement.
Beyond that, we have our performance and compliance metrics that really are how we differentiate one vendor to the next, in terms of their outcomes, you know, beyond the financial. And with that, you know, I think if they’re performing at their best, it’s going to translate, you know, to those outcomes that we need them to see, which will result in, you know, increased market share.
So, yeah, I think we in the industry are seeing an expanding pie. You know, I don’t think that that’s at all unusual for most of the folks in our space, but how do you carve that pie out, you know, in terms of market share, I think comes down to how these folks are performing in compliance and performance and it should translate. So in the best of all situations, we’re having conversations about the financials, we’re having conversations about their strategy, which would be inclusive of things like digital and AI.
And we’re seeing those outcomes in the day-to-day metrics that we measure, or then feeding back to the clients and saying, these vendors are performing great and we’re rewarding them with more business. These vendors are not performing so great and they’re the ones that are having business taken away.
Adam Parks (25:20)
We are watching a lot of the same things. You’re just expecting a change in those measurements based on whether it be that reduction in cost or that increase in productivity or both depending on the use cases that vendor or organization is is actively deploying.
Pete Klipa (25:38)
Yeah, I mean, it’s a results review, but it’s also coupled with a dialogue. The dialogue is around that strategic push. We tend to want to challenge our vendors a little bit about what the future looks like for them. ⁓ One of the things that we know is the case in the law firm space, because there’s a lot of folks who are more gray beard oriented like myself. I put myself in that category.
And there’s succession planning and things like that that are discussed as a matter of routine. And as we have those conversations, it’s a conversation about the future, but it’s a good opportunity to have a dialogue about what the future initiatives of the firm or the agency look like. And that goes beyond the people side. That’s the commitment to technology and some of these tools and enhancements.
You know, it might be a system migration. You know, that’s something that many folks consider. And, you know, in the course of that system migration, especially with the AI consideration around programmatic advantages and being able to program quicker and customize in mass fashion quicker, you know, those things are things that can be differentiated, you know, as it relates to changes, you know, at vendor providers.
So we want to understand what is their thinking on that today, as well as kind of in the short to intermediate term, you know, how’s that going to change, you know, for us? And ultimately, what do they expect to happen from it? You know, because, you know, all this AI stuff, it’s important, but if it doesn’t translate to results that will be beneficial, you know, then I think it somewhat misses the mark.
Adam Parks (27:24)
I think organ-, I think some organizations are struggling with how to truly measure the return on investment for the money that they’re putting into these AI tools. And it’ll take us some time, I think, before we truly understand it and can isolate it. And I don’t know that everything can be isolated.
But you bring up some really good points and and how, and some perspective for our audience that are those vendors that are being looked at by their clients and some of the things that you may find to be more important than others, because I don’t think it’s always clear cut. And although a lot of clients may be asking for the same things, they may be interpreting them somewhat differently.
So Pete, I really do appreciate you coming on sharing all of your experience and insights. This has been a great conversation.
Pete Klipa (28:11)
My pleasure. As always, you guys do so much to further the industry and share knowledge amongst a very large group across our space. And we always look forward to these updates that you guys have.
Adam Parks (28:26)
Well, I really do appreciate you. For those of you that are watching, if you have additional questions you’d like to ask Peter 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 in the comments below as well.
And hopefully I can get Pete back here at least one more time to help me continue to create great content for a great industry. But until next time, Pete, thank you so much. I appreciate all your insights.
Pete Klipa (28:49)
Thank you, appreciate it, and as always.
Adam Parks (28:51)
And thank you everybody for watching. We appreciate your time and attention. See y’all again soon. Bye everyone.
