Adam Parks (00:00)
Hello everybody, Adam Parks here with another Receivables webinar. Today we're here talking about really challenging the conventions around CRM systems, communication orchestration, and artificial intelligence and what that intersection looks like.
So we've got some great speakers with me here today, kind of going around, I guess, in the order that you are on my screen. Anshul, starting with you, can you tell everyone a little about yourself and how you got to the seat that you're in today?
Anshul (00:30)
Yeah, thanks, Adam.
Hello, everyone. My name is Anshul. I'm the co-founder and CEO of Vodex.ai. At Vodex, we are building an AI coworker for the collections team. Basically, it's a kind of system of engagement where the AI agent interacts with your data not only over voice AI, but also over other channels like SMS and email. And it does a fully contextual conversation across multiple channels. So yeah, just building that, working closely with a lot of collection agencies. That's about it.
Adam Parks (00:58)
Fantastic, fantastic. And Ian, how about you? Could you tell everyone a little about yourself?
Ian (01:02)
Sure, I'm CEO and co-founder of Lateral Technology. We are a modern collection platform with built-in communication channels around email, SMS, and Letter, and we have a plug-in architecture that can integrate with AI vendors such as Vodex.
Adam Parks (01:20)
Fantastic. And Daryl, how about you? Could you tell everyone a little about yourself and how you got to the seat that you're in today?
Darryl Brown (01:24)
Yes, my name is Darryl. The owner of Greystone and Associates. We are a collection agency. We work closely with Anshul on his software with Vodex as well. So yeah.
Adam Parks (01:37)
Fantastic. And so gentlemen, as we talk about CRM systems in general and the challenges that we have when it comes to CRMs, talk to me a little about some of the challenges that you are facing as businesses and how you've started to bridge that gap between the CRM and the let's say the communication technology.
I'm gonna float it out there to anybody.
Darryl Brown (02:02)
Okay. I'll start it off. Mainly the CRM is kind of the focus point. We need a CRM to be able to scale our business. It's very important.
Just as of recently with a lot of this new technology, AI and a lot of things that are coming about, we're trying to figure out a way to improve the collection experience. That's why we partnered with like I say, Vodex, Dros, and a lot of other companies to kind of create an all in one collection platform, just to streamline the data collection experience.
Adam Parks (02:35)
And so you were looking at these fragmented systems, you wanted to be able to pull this data back together again and kind of start thinking our way through the next best action. But it sounds like, I mean this from a lot of organizations I've talked with, that companies have been trying to piece these different things together.
What drove you towards using a holistically designed system versus piecing and comparing something together to try and solve the same problem? Like what was it that drove you in that direction?
Darryl Brown (03:06)
Mainly collections is more of a numbers game if you wanna get through the accounts in a timely fashion to increase your liquidation. At the time we were working with a bunch of different companies. You know, we had a company for text, a company for email. Sorry about that.
Yeah. Sorry about that guys. We had a company we were using for text, a company for email, a company for C R And yeah, we kinda just wanted to create something that's all in one.
Darryl Brown (03:34)
Like I said, just streamline the day to day collection process. Also as of recent years there've been a lot of changes with AI and just technology in general. So we wanted to kinda improve our tools, so to speak, based on what was out there being offered currently.
Adam Parks (03:52)
Okay.
Now Ian from your perspective, right? Like you've been working with a lot of different groups from a CRM perspective. What made you look at it and say, Okay, hey, I want to tie up to another piece of technology here? Is it a focus on that core piece? Like what brought about a partnership and why would you look at third party solutions?
Ian (04:13)
Well, my philosophy is that you can't be the best at everything. And I think CRMs that try to do everything themselves end up falling short in different places. So where lateral is strong is in flexibility and our plug-in architecture, which makes it great for the fact that we can then use our plug-in architecture and our flexibility to plug in pieces of other vendors that are really strong in a seamless kind of fashion.
So people aren't exporting data from the CRM and then importing it to another system, right? And creating that extra overhead. Lateral is built so because it's cloud based, everything is API driven, we can have that seamless connection with other systems.
So because of what Daryl just said, the pace of technology is changing so fast, there's so many new competitors that are coming into the space now.
One of the risks for a software company such as lateral is that AI will level the playing field, right? And everybody can try to build a CRM. But the reality is that no two agencies are the same. So for example, there's legal collections, there's the legal workflows, there's compliance rules, and everybody needs a little bit of a different piece of the puzzle.
So again, I just to reiterate, I think it's more and more difficult to be the best at everything given the pace of change has sped up so fast using AI technology that software companies such as Lateral really need to focus on a core strength. And we decided our core strength is going to be in having the best, you know, core CRM platform, the most flexible, and then we'll let
People like Anshul and Vodex really hone in on what they want to be best at.
Adam Parks (06:07)
On the smaller pieces and Anshul from our conversations, you've talked a lot about being able to bring context across the channels. And so how does that kind of fit into the equation here?
Anshul (06:20)
So like what Daryll has mentioned and what Ian has mentioned CRM is really, really important.
piece of technology in collections, especially now. The number of accounts is growing. The number of channels are increasing. So that's where we specialize. So we started focusing on collections. So when we started, we were just building voice AI agents, which can make phone calls, make outbound calls, and receive incoming calls. And we started focusing on collections. So we really trained our AI agent to understand the compliances, the regulations,
the reg F, you know the seven and seven rules, all those things. But what we realized is that just voice agents are not sufficient. If an AI agent is making a call, it's not about just making a call, right? It's about the contextual conversation across what is happening across multiple channels, what is happening in the CRM. So for example, let's say a debtor sends an SMS or an email saying that I lost my job or you know like I won't be able to make a payment. Does that context pass to the AI agent? Does the AI agent know about what is happening over other channels? So that is very important. What was the conversation that has happened with the data in the past, last month, two months back?
So that context is very important for an AI agent. And when that context is passed to the AI agent, then the conversation becomes natural, more effective. So what we say is like, you don't just need the system of record. Basically, the CRM is your system of record. But you need a system of engagement also, which pulls the information from your system of record and does a really good job of engaging with your debtors across multiple channels in a fully contextual way. That makes me very human, like experience you know.
Adam Parks (08:10)
And Daryll, from your perspective, you know, how are you doing some of that prior to these types of tool sets?
Darryl Brown (08:16)
Just to kind of pick it back off what Ian and Anshul was saying, technology has changed over the recent years. So we had previously we're you we're using like outdated tools so to speak, broadcast dialer, but now we have a more intelligent dialer with AI voice agents.
Yeah, just improving what tools that we were already using. We just kind of work with Anshul to revisit and find ways to make those better. whether it's you know, our financial dashboard to kinda see our stats, working with other vendors for text messaging.
and show for the AI agent calls. Just trying to have everything in one space to kind of again just streamline the day-to-day work workload. So yeah, that's what we've been doing.
Adam Parks (09:03)
Understood. And I'm curious about how this has kind of evolved over time. But you know, we're talking about a lot of different channels here and you know, Ian from your perspective, you've got people using all of these different tools. How do you carry context across a conversation that hits multiple channels. Right? Like is the CRM kind of stuck in that orchestration position in between all of these other channels? Is that still an API driven thing or like how does that actually function?
Ian (09:30)
Well, I mean where does the context need to sit is the first, you know question, right? Yeah, so from our perspective, I mean we run this actually quite recently where we have a vendor, I won't mention any names, that they like they also so our system ha usually uses different vendors for letter, SMS, and email.
Adam Parks (09:35)
Good question.
Ian (09:54)
We have another vendor that does all three and they say, Hey, you know, take a file, we'll handle it for you. And then the challenge with that is getting the information back. So when we send an email via API, right through our system, we know if it's opened, clicked, bounced, all of that data goes right into the CRM. So then we can make decisions like if the email is bounced, send a text message.
Ian (10:20)
So we believe the data should sit in the CRM because that gives the CRM the ability to make those decisions right in the CRM. And some clients, larger clients will use a data warehouse where we have to send you known data to a data warehouse and then they're gonna hire their own AI people.
So that's another strategy. I think there's no single answer, but from my perspective, especially for the small to medium sized agencies, the more data in the CRM, the better, because then there's a single point of truth that you can use AI the most effectively from.
Adam Parks (10:58)
I mean we got another you know words that I use for CRM which is Sisma record. And so internally I call it an SOR. Like that is the location which is the master management of accounts. Maybe it's more of a banking term, but I look at this very much the same way. But Anshul when you started building a tool set like this
What were you seeing in the market and how were they moving that context prior to having an orchestration layer or some way to consolidate this data back to the CRM itself?
Anshul (11:33)
So as Ian mentioned, the context is very important. And also we need some place where, you know, like the source of truth should be there. It should be somewhere like CRM should be your source of truth and all the information, all the context, you know, what kind of conversation is happening across multiple channels that should be there in the CRM.
So that is very important because then you can go to that CRM, you can fetch that information, you can write that information back into that CRM. So that is very important to create a unified conversation across multiple channels because there will be multiple channels right now. There is SMS, email, there is call, there is live chat.
You know in the future you will have LinkedIn and you will never know about Snapchat, Instagram.
Adam Parks (12:25)
Well you've seen WhatsApp in other countries. I don't think it's crazy to think that we'll communicate through other channels in the not too distant future.
Anshul (12:31)
Yeah, but the thing is, there should be a single place of source of truth where you can bring all that conversation in a single place. And then that information can be passed to your AI agent, to your human agent. They can see a single place. Earlier, what we have seen is, in some of the agencies, they were using spreadsheets also to manage the context, like tracking information in spreadsheets. There are a lot of CRMs in the market.
Anshul (13:00)
The thing is, those CRMs were not properly integrated with other channels. Like for example, the AI calls. the AI calling, for that matter, even the human calling tool, that was also separate, not properly integrated with the CRM. So a lot of context was getting lost. You are using one tool to make the phone call. You are using another tool to send SMS. And those tools are not integrated with your CRM. So some communication is
happening over your human calls, some communication is happening over SMS, some communication is happening over AI calls, and those tools are not really integrated. Because of that, a lot of context was getting lost, and I saw that as a big problem. You cannot build a really good voice agent until you have all the context unified in a single place. So that was the main motto, like we need to have everything in one place, and that's why we started
started saying that we build a contextual voice agent, not just an AI agent which makes phone calls, but an AI agent which understands the context across multiple channels. Ultimately, it's the point. The point is that how do we make the life of a collector easy? The collector should not be relying on their memory or their brain to recall everything. The CRM should be the source of truth, and it should just tell the collector that this is the conversation that has happened.
Anshul (14:24)
over SMS, email, or any other channel. And the collector should be able to just take an instant decision and take action on that particular account without worrying too much about what happened over multiple channels. So yeah, that's why context is really important.
Adam Parks (14:39)
it makes me think a lot about the context perspective, how do we capture context from a live collector call?
And for years it's been, okay, we're gonna take collector notes, but then I've got two hundred collectors, they've all got their own shorthand. I've got all these different languages that are, you know, kinda let's call it sub languages that are happening within my notes, which makes it very unstructured and a lot more of a token burn for me to try to evaluate or understand. You know, when when we think about it from that perspective, you know
What has your experience been in trying to bring new context back to old phone calls? There's a lot of gold in their collector notes. So I'm really kind of curious how all three of you have addressed the collector note issue.
Anshul (15:26)
AI is your answer. AI can do wonders.
Adam Parks (15:30)
Yeah, look at but look AI is not just the magic bullet, right? So it's yes, there's a lot to it, but it's not just the magic bullet and yes, there are some things there, but for so you're saying you're gonna run it through an AI model and structure that data for querying, or like what's that what's that look like from a technical perspective?
Anshul (15:49)
So basically we go through the transcripts and we run through that model. There are a lot of models. There are some open source models also which you can deploy. So you don't need to worry about your data going to OpenAI or Anthropic. So you can just pass on that transcript and then you can extract the meaningful information from that human call also. There are some important insights that you can fetch from those calls.
For example, you can literally use your small language model as a LLM as a judge. Basically, what you say is like, go through this transcript and tell me whether this person is having any kind of hardship, if this person is willing to make a payment, if this person is promising to make a payment. You can ask all those sorts of questions and you can extract very structured data and you can put that data back into the CRM. LLM makes all those things really, really easy.
Adam Parks (16:45)
And Ian, from your perspective, right, managing a large CRM, how have you looked at or addressed or seen clients addressing that challenge of driving context from that unstructured data?
Ian (16:56)
Well, I'm gonna take another approach and kind of talk about it from a return on investment perspective. So I think agencies invest based on how many collectors they have, how much data, what value is that incremental collection rate worth and how much money am I willing to spend
Adam Parks (17:05)
Okay.
Ian (17:20)
to push that 1% or a half percent up. The reality is, you know, there's been things that you know collection agencies could be doing for years and didn't do. You know, so for example, you know, we have workflow, we've had workflow automation for 16 years. Very few of our clients ever bothered doing A-B testing. You know, they never bothered
Creating a champion challenger workflow to see which one performs better. Why? Well, because there's a cost involved. There's time and effort, splitting it into two workflows, seeing which one performs better, you know, having just the time and energy to do that, even though there's not a big cost, you know, there's opportunity cost. And so I think when it comes to AI.
It's really nothing, it's just a new perspective and maybe it's getting a lot of hype right now because of all the great AI tools. But from a pure data science perspective, you know, there's things that people could have been doing 15 years ago, 10 years ago, five years ago, and they're just thinking about doing those things now because of all the hype. But, you know, there's lots of things that you can do to improve your collection rates that don't involve an LLM.
Adam Parks (18:14)
Sure.
Ian (18:34)r
you know Claude or ChatGPT or any of these tools, right? so I think it's ultimately about ROI. you know, how much do you want to spend to increase your collections?
Adam Parks (18:39)
Sure. You can just bring even just looking at it from an organization and orchestra, I would think that LLM aside or AI aside, that there would be significant value to unlock in
Coordinating your data silos in and of itself. So just bringing the data together, I think, is that first big step. As someone who's gone through a lot of these system conversions through the years, what kind of impact have you seen on organizations as they have brought the data together and can now learn across a larger data set?
Ian (19:18)
That's a good question. For me personally, you know, you can go buy the best Ferrari, but if you don't take driving lessons, you know, you're not gonna be any better driver than anybody else. So it's not just about the tool that you buy, it's about how you decide to use that tool. And the same is true with the CRM, like and the data inside it, you know.
Ian (19:41)
There's very few companies that are taking a look at all of the contextual data and doing something with it. So whether or not you have that data, I don't think that's the question. It's you know, people aren't even bothering to do something with the data that they have. Again, it's a time and effort equation.
Adam Parks (20:02)
Can they activate the value of the data that they have today? And if you can't activate that data, what's the point of collecting a thousand other lines of data? Is that is that one of the
Ian (20:10)
Right. And how much are you
willing to yeah, and how much are you willing to spend to prove how much value you can get and how much incremental improvement. I think part of it is because and this is where, you know, you have different departments that try to own their space. The head of operations thinks they understand collections and they do it a certain way. They've been doing that way for fifteen years, they know
You know, hey, we call three times and then we send two text messages and that works. You know. Then you got the tech guys saying, no, you know, we gotta have AI, we gotta do this, we gotta we gotta do that, we gotta we need a but a huge budget, you know, and and so these different competing departments in a collection agency, and and keep in mind a small collection agency isn't gonna have an IT team, you know, so they're gonna rely on people like Anshul or myself to advise them.
Adam Parks (20:44)
That doesn't work.
Ian (21:02)
So this is a complicated equation and I think it's related not so much to what's possible, but how much does it cost and what are you trying to unlock? You know, are you trying to improve your collection rate? And if so, how much are you willing to pay, you know, somebody like me or Anshul to improve your collection rate? Because ultimately, you know, all of this technology costs money.
Adam Parks (21:02)
Sure.
No, no question there.
Anshul (21:28)
Yeah, I want to add something here. So what I mentioned is very true. The workflows are really, really important. It's not about the CRM. You can have a really good CRM. But if you don't build those workflows properly, you will not be able to capture the full potential of the CRM. And the good thing with AI is it helps you make those workflows really, really easily. And it can automate a lot of those workflows also. You can quickly build in those workflows. You can automate some of those workflows.
That's why this time it's different, you know, and more and more collection agencies will adopt, you know, these workflows. They will implement these workflows because they don't need to spend too much time. They can literally develop these workflows in minutes rather than days, you know. So how good are your workflows? That's very important.
Adam Parks (22:14)
Yeah.
Ian (22:17)
Yeah, yeah, but I'm just gonna challenge you a little bit there, Anshul If you're talking about a collection agency with that's been just has one workflow, right, and they've been doing it that way for fifteen years, and you come along and you say, Hey Ian, I've got this AI, you know, I wanna plug in. If I'm the IT guy, you know, and you know, I happen to have a computer science degree, you know, I'll say, Well, Anshul there's no data, your AI isn't gonna be able to learn
So what are you gonna do? You're gonna reference other people's data and then run it against mine and make it get very bad. This gets very muddy because the AI is usually only as good as the data that it's analyzing. So if you don't have a lot of different workflows and you haven't been sending emails, you haven't been sending text messages and all of this digital communication is new, there's nothing to train the AI on, you know, quite frankly.
Adam Parks (23:10)
There's no differentiation.
Yeah.
Anshul (23:12)
Yeah. But I would say better late than never. Like, they can start now.
Ian (23:13)
Right. Yeah.
Adam Parks (23:17)
I think the first, the first piece of any good I'm looking for, I'm going through a big data project internally here where I looked across the entire media organization and said, like, where's our data live? Right. Found thirty-two different locations, which equated to about forty two hundred signals across all those data sets. But before I could sit down and say, What should the dashboard look like? The first thing I had to understand was like what data do I have and where does it live? And once I get past that series of challenges, then I can start to address okay.
How do I want to visualize?
webinar data, podcast data, whatever. And then I could start kind of working away through it. Now a lot of the project that I'm doing there is being driven using artificial intelligence tools, because how long would it take me weeks to put together a spreadsheet of 4200 signals, right? Like that would have been a big long process. But by sending a bot down the API paths and being able to pull back that structure and organize it for myself, you know, it turned into days, not weeks or months in order to execute the project. I think about the
application of artificial intelligence to the debt collection industry very much the same way in that where the data exist today? What's coming together and what's happening? So Darrell, when we first started talking, you had mentioned that you had you know come from using all of these different vendors and tools for different things. How different has life become as those things start to get combined?
Darryl Brown (24:40)
yeah. Ian and Anshul both made pretty good points.
If we step back like the previous time before AI, you know as Ian said, a lot of these companies are kinda you know, they found out what works for them and they just kinda ran with that. but workflow is very important because
Before AI, it relied solely on the collector to remember, hey, I gotta give this person a call back, or hey, I might need to call this person two times a day, or I might need to send them a text or email. If you have that workflow in place, it will kinda is giving everybody
a better opportunity, I guess, or it's kinda leveling out the playing field a little bit where you don't I mean, a collector can be good, but if he has a great CRM and a great workflow and he can that can kind of organize everything for them, it'll tell them what to do day to day and that will in return increase their collection rate.
So right now it's kinda like we're bridging the gap between, you know owners like myself who've been doing things one way for so long and then we have, you know, guys like Ian and Anshul who are great in technology. So we're just trying to bring it together and you know, to improve the business and see what works, what doesn't. The only way you can do that is, you know, trial and error, create workflows and see what works and see what doesn't work. But it's been been a fun opportunity. So yeah.
Adam Parks (26:02)
What's the experience look like for the consumer? So we've gone through moving, we're creating these new channels in new contexts, but what is that equated to in terms of new friction for the consumers themselves? Have we seen any change there? Are you seeing a higher engagement rate? Like, is there anything that you can point your finger at and be like, hey, I'm seeing this move in my KPIs now that I've started to bring that together?
Darryl Brown (26:27)
I think it's a combination of everything. Some people might respond better with text messaging. Just an example, I had a young lady this morning. She works at the hospital.
She's like, hey, you know, between patients, I can respond to email. They're not wanting to sit on the phone all day and have long drawn out conversations. So that's where email comes into play. text messaging. Everyone is different. so you have to have all the two tools to maximize the collection.
Like I said, I think it's important to utilize all of those tools and create workflows and just see what works because everyone is not, you know, everyone is different. So it's important to ⁓ have those different avenues.
Adam Parks (27:07)
Anshul, any thoughts on the reduction of consumer friction?
Anshul (27:13)
Yeah, so.
Bidding on what Daryll said, I think it's very important, and that's the responsibility of a good CRM, to identify the best channel on which you should reach out to a debtor. You should not just blast SMS or blast AI calls to all your debtors. The system should tell you, basically. And that's the importance, and that's the benefit of capturing the data. You will get to know who are the consumers who are more responsive on SMS, who are the consumers who are more responsible over email. And another benefit is propensity. You can identify which data have more propensity to pay, who are more willing to pay, versus who are having some kind of hardship or trying to raise a dispute. So that way you can, you
channelize your energy on those people who are more willing to pay. And you can reach out to them through the right channel. And basically, you can have a better experience for your consumers. So that's why I think that's the responsibility of a good CRM, that it tells you which channels to use for which debtor, who are the debtors whom you should reach out first, the easy to handle, persons difficult to handle, those kinds of things. And that can be done only if you're capturing the data.
Adam Parks (28:30)
So one of the things I wanna point out and just kind of discuss here a little bit is the difference between AI in the common sense in the industry. We start talking about voice bots and I feel like the whole world has gone VoiceBot crazy recently and like this is the focal point in their mind.
But from what we've seen in other countries and of the academic reports that are available related to call center and AI technology, what I'm seeing is that the biggest impact use case is actually the deterministic workflows, which account is going down which channel for which reason at which time.
etc. And that piece is that larger piece. The second largest piece that I was finding in those reports was that the co-pilot style, meaning empowering and activating the collectors themselves, was like one of the next most powerful things that we could be doing. Putting the right tools in place to get the live collector or empower them to make the best possible decisions, to negotiate, to do whatever it is that they need to do for that purpose. You know, as we kind of start thinking about it from that perspective, any thoughts on what that's gonna mean?
Anshul (29:47)
Yeah, I mean, that's a really good use case because you know, when a collector is on a call with a debtor, there are so many things you have to consider, you know, what is his paying pattern, how this person has been paying, what is like the credit bureau data saying about this person, you know, or what conversation has been happened with this person in the past, like six months or in last one or two years.
And it is very difficult for a human agent to simultaneously talk also with a debtor and also go through different systems and screens and figure out and find and search all that data. So that's where AI can help. AI can automatically listen to the live conversation. And based on the conversation, it can surface that really important information in front of the live collector. So the collector can just see on the screen what is relevant for this call. And yeah, that is a really good use case, I would say. But yeah, happy to hear what Ian has to say in this order.
Adam Parks (30:45)
Any thoughts, guys?
Ian (30:47)
I'll jump in. I think there's a lot of AI use cases, so it really again depends on KPIs that you're looking for. So for example, a collection agency that wants to minimize their cost of collections might say, right, I want to voice AI, you know, as many voice AI agents as possible. Another agency that says we want to increase our collection rates because we have clients that are on a scorecard.
And we need to keep those clients, you know, an assisted agent is gonna probably perform better. So using an AI assisting humans, I think is gonna perform better than a pure AI bot personally. So I think again, it's about what tools do you have at your disposal based on what KPIs are you trying to improve today?
And you need to really talk to people like me and Anshul to help answer that question because it's really, you know, that is complex. And, you know, as much as you know, we love AI, you know, that's not very helpful for somebody who's starting a new collection agency from scratch, right? Because, you know, they have no data, there's no data room. So we need to have tools that work out of the box without any AI, without any data.
Right. And that means you have to have workflows and you have to have some human intelligence. An AI agent will never say, Hey, what color is the envelope that you send your letters? You know, an AI agent won't ask that question. A human will. A human might say, Hey, let's try changing our envelope color to red, you know. An AI agent probably won't come up with that suggestion.
So, you know I am maybe a little bit different from Anshul. I don't think the answer is AI. The answer is based on how you are framing the question, you know? How you frame the question will determine, you know, what answer we need to give. And the answer isn't always, ⁓ yeah, we'll solve it with AI, right? That's not always the answer. That's my opinion.
Adam Parks (32:47)
Well that's why I talk about the co-pilot, right? It's about empowering that human decision. If you look at the anthropic report from back in March related to the job markets, you can see that there's no observed loss of jobs happening right now.
Like some of the things that we're envisioning in this new AI driven world are are not coming to fruition the way that I think people were expecting it to happen because they thought that all these jobs would get replaced and all these things, but there's been no observable job loss within the theoretical ⁓ AI impact areas, right? There's a great report that Anthropic released back in March and it goes into some pretty significant detail there. But then as I as I started going through other you know, use case tools and looking at it, look there's a million use cases for artificial intelligence, specifically related to debt collection, I could categorize them into seven. Really six. And I think the seventh is this new idea of the agency that's being built from the ground up, specifically empowered by artificial intelligence, because they don't have the
data context challenges, right, that existed five, ten years ago. Because even if you think about what's available today versus what was available six months ago, eight months ago, and you think about if I built a business eight months ago, what would it look like? It would be very different if I started building it today. So what's that going to mean over the next six, eight, twelve months? But Ian to your point,
I think people chase the next shiny thing and I don't know how long it's gonna be shiny if they can't support the cost and investments with a revenue stream. So it's gonna have to work in order for people to continue to make that investment because debt collectors are, you know, shrewd business folks and they're not going to spend unless there's a solution or some sort of an increase to that revenue.
Daryl, from your per you the you're the collector, or the current collector in the room here, right? Any thoughts on how you look at the challenges?
Darryl Brown (34:51)
Yeah, I, me and Anshul talk about this a lot and I think AI is never gonna replace the collector, because to be a great collector, you would have to have some type of empathy. You have to relate to the consumer. and that's just something a robot can never do.
I think AI is more so should be used as a tool to assist our day-to-day workload. So that's the main approach we take with using AI is just finding ways that it can, you know, maybe help streamline or speed up our process that we have day-to-day. Like you said, I mean just basic you know, workflow that we use every day, AI, chat GPT can kinda speed that process up.
Adam Parks (35:34)
I think that's the next set of use cases. And when I say that it's seven specific to debt collection, I do mean like in our debt collection world, like you know sourcing of pending data, ⁓ negotiation, voice, there we've got those specific use cases.
But I think there's a lot of opportunity for deploying artificial intelligence across our organizations because what we don't talk about in the debt collection industry is the ability to leverage AI tools from an accounting perspective, HR perspective, recruiting retention perspective. Like there's all these others I'm gonna call them general business use cases that are not right. Like HR and accounting is not specific to the debt collection industry. That's a general business need. But sourcing the pending data files and
Scrubbing files is definitely a debt collection, you know, specific activity. And so I'm starting to think my way through like what is this gonna look like over the coming years? So as we go into kind of our final quarter of our discussion today, guys, what do you think happens over the next man? I was gonna say next five years, but I feel like I should probably ask like, what do think happens over the next six months is probably a better tactical question for our audience today. So, you know, what guys, I would kind of float it out there for discussion.
But what happens next over the next six to twelve months?
Anshul (36:51)
Yeah, well, I think collection agencies and you know, I have been talking to a lot of collection agencies since last more than two years and like the discussions that we had two years back
is very different from the discussions we're having now. I know, earlier, like two years back, they were very hesitant to implement AI and they were like, I'm not interested in implementing voiceboards and AI right now, but now things are changing. They want to implement, but they want to start implementing some low hanging fruits, now, some easy to implement use cases. And one of the easiest to implement use cases that I come across is the inbound call handling that to after office hours.
So let's say, for debtor calls after 9 p.m. at 10 p.m. or 2 a.m. in the morning, there is no debtor, there is no collector who is available to talk to the debtor. So you are anyway gonna miss that debtor. Maybe that person has intent to pay, but there is no one available to talk, so you miss that revenue. There is a revenue leakage, but that can be fixed by implementing an AI agent, a voice agent, which can handle some of the calls after office.
If the person is looking for like, why am I charged for this particular thing? Or how much do I owe? Or give me the payment link, I will make a payment right away. AI can handle those easy use cases. What I think is in the next six to 12 months, collection agencies will start with implementing some of these easy to implement use cases. And then once they're more comfortable with the compliance, the regulation, they are seeing the ROI on their investment, then maybe they will go towards with more advanced use cases. Yep.
Adam Parks (38:27)
Interesting. Interesting. Daryl, you know, from your perspective, how do you think things continue to evolve over the next six to twelve months?
Darryl Brown (38:34)
Just to kind of piggyback on what Ansha was saying, I think again, AI is never gonna fully replace the human experience, but I think we can use it as a tool to just improve our experience. So yeah, that's pretty much it.
Adam Parks (38:49)
And Ian, how about you? What do you think is coming down the pipe in the next six to twelve months?
Ian (38:53)
Well this might not be the best sales pitch, but I actually think it's about implementing AI in other parts of your organization operationally is gonna be kind of the next thing, kinda like what you're saying, Adam. People are starting to use Chat GPT. It's like, well, how can we just make communication more effective? How can we make you know,
I need a revenue forecast for the next three months. I need to it's how we're using Chat GPT and these tools every day, how we integrate that into our business life. I think that's what's happening now.
And there's more and more tools coming that are, you know, I'm actually beta testing a brand new tool, you know, that helps, you know, lateral with, you know, finance and HR and putting everything together in one place. You know, it's not even a tool that Lateral built. So these tools that I'm you know that I'm kind of excited about 'cause I'm using, I think other people will be using and just our usage of AI is going to be increasing so high that's what I see happening over the next six months.
How that affects Lateral, I'm not sure. You know, but I think everybody's gonna be demanding a lot more from technology companies, that's for sure, you know, ⁓ to so you know, we'll probably be busy.
Adam Parks (40:14)
Do you think that we'll start to see more of, let's say, an increase in how we are or a change in how we access, visualize, and understand the data that we do have today?
If the last 12 months have been all about getting that data into structure, do you think maybe over the next six to twelve months we might start seeing more organizations with a focus on Chatbot reporting, you know, question visualization like that. As people have become more comfortable with prompting and prompt engineering, do you think that that becomes a more common phrase, or do you think that the report building and understanding of that data remains in the IT department as a series of reports? Or do you think we're gonna start delving into the data a little deeper?
Ian (40:56)
Well, I this is where I completely agree with Anshul where one you know, two of the best features that a CRM like lateral can offer or that you can add to your CRM, however you want to think about it, are you know, next best action, propensity to pay score, and the amount of agencies actually, you know, that are funding those things is very small. Partly because, you know, how does the CRM interact with that information?
And you know, the cost of that has been high. So the cost of AI is going down. So this again becomes a money equation where the cost is going down. So more and more agencies are going to have those types of tools. And I think there is going to be a big, huge ROI benefit for getting those specific data points into your CRM and making decisions about those that agencies are going to wanna start unlocking very quickly in the next six to twelve months.
Adam Parks (41:53)
Interesting. Anshul how about from your perspective? Any insights as it relates to reporting and understanding the data at hand?
Anshul (42:01)
I mean, yeah, people are very familiar now with ChatGPT and Cloud and all these tools.
I think this chat interface will become default going forward in the next six to 12 months. So people will expect this chat kind of interface in all their software, like in their CRM and any other tool. What they want is like, nobody wants to ask their IT team to give me a report. Think of a collector or a collection agency owner. They want to just chat with the application and get all their data back. So if they can chat with their CRM and if they can get their data back in a nice, easy to visualize format. And I think it's a no-brainer. Why would anyone not want that?
Adam Parks (42:43)
Yeah, I mean I definitely see it happening because even from a proof of concept perspective, right? Like I don't do a whole lot of you know coding or anything anymore, but I'll spend a lot of my time in perplexity or claude or whatever tool working through what do I want to see. Right? How do I want it to look? And then I can, you know, I can fight with the chatbot about, you know, making that dot making that button green or whatever, and I'm not in this back and forth with the dev team.
So I feel like it at least empowered me to be able to better formulate what it is that I want. You know, it's like decision paralysis. If I got forty two hundred signals in front of me, where do you even begin? Right? It's the same reason McDonald's used to have 26 items on the dollar menu and now there's eight. Right? It's decision paralysis. People pull up to the thing they can't decide what to do next.
But being able to go through that exercise without dragging IT team down and burning all of that, not just my own brain power and when I want to do it, but like when you actually have to drag a whole team along with you, I think it changes the dynamic of that pretty significantly.
So Joe, and with all of that being said, when we're thinking about the AI tools that organizations are rolling out now. We've talked about the reporting aspect of it, the visualization. Do you think that and we've talked about reporting from the perspective of the operator themselves, but do you think that client expectations change from a reporting perspective over the coming years?
Are the clients gonna now have a higher level of expectation for direct connection or the ability to query information right now that we're not having these series of SQL professionals in between you know two parties.
Anshul (44:25)
Yeah, absolutely. I mean, yeah, it's everyone's expectation now to make it super easy to query the information they want. So it's basically people like us, it's our responsibility to build those systems, like people like Ian. I mean, anyone who is building in this CRM space, should be able to, like, there are a lot of tools nowadays for developers, right? Like we have tools like Cursor and Bolt and.
and probably cloud code, they are giving a really good experience to the developers. So it's our responsibility to give that kind of experience to the people in collections, ecosystem collections domain. People who are clients and collectors should be able to just chat with their application and get their data back. So yeah, definitely people need that. We should be able to.
Ian (45:15)
I kind of disagree. I mean, I think it's not necessarily the CRM's responsibility to do that. The CRM might have to give up its data though, right? The answers are in the data. I mean, we currently integrate with a BI tool called Metabase, right? And people can use Metabase. Now there's lots of BI tools not coming now where you can just talk to it. And we're trying to decide if that should be in lateral or outside of lateral.
Ian (45:41)
And you can make arguments on both sides. So yeah, we'll probably be gonna have that feature, but there's so many BI tools that are also having that feature that are gonna be doing the same thing. So I'm not actually sure whose responsibility it is, but to your point, Adam, I definitely think clients are gonna be demanding access to their data so that they can be doing those things.
Adam Parks (46:07)
I agree. I think they're gonna demand access to the data and I think they're gonna demand some sort of a querying functionality, whether that's a chatbot or whatever, or a visualization studio, whether that happens inside of the CRM or through a partnership or data you know, or business intelligence tool. I don't know the answer to the mechanics of how that will execute, but I could definitely hear the rumblings of people going.
Well, if I can ask this to Chat GPT, and now that they're understanding the level of context that can be brought together for that type of query and you know how can we organize that and if look if you had asked me a year ago you know is ChatGPT going to be on the decline you know in 2026 I would have laughed and like my god like how could they possibly lose steam now? And we've seen everything shift over to Claude but as if we've seen the shift to Claude and perplexity in other tools we're starting to see these opportunities to visualize in new and interesting ways. And even folks with zero coding capability now can build proof of concept in perplexity computers.
Or cloud code, whatever, and they can start working their way through that stuff with no experience with almost no computer experience. Perplexity, probably the easiest, at least for me. And now once you can start visualizing it, though, you go down these rabbit holes. Like now you're excited because you can activate the data that's been dormant in your systems for generations.
Anshul (47:17)
One thing though, one quick point I would like to add here, the demand has always been there. The client has always been wanting to see their data. But historically, it was not that easy. You need to go through BI tools, and it was not that easy to operate. But now with all these AI systems, it has become much easier to get the data that the client wants.
Adam Parks (47:37)
Sure.
Ian (47:45)
I mean I hate to be the guy that brings up compliance and security, but you know, there's real compliance and security concerns. So there's performance concerns. So for example, our larger clients that have a BI tool, we put on a copy of the database, which is three seconds behind the live database, because you don't want people querying your live database and making some stupid query and it slows down the experience for everybody else, right?
So there's lots of considerations and this is PII data, you know, you're talking like, you know, people's social security numbers and all of this. So there's a lot that has to go into it. We can't just hook up, you know, Claude and Chat GPT to our softwares and say, Yeah, here you go, client. You know, it's a little bit more yeah, it's a little bit more nuanced, you know, than that.
Adam Parks (48:34)
Well it's significantly more nuanced, right? There's rappers involved. There's I mean, if we could go down the technical discussion, I think there's a lot of opportunity here. But I think the the challenge is always the compliance pieces, but I don't even think that the compliance pieces are going to be
the biggest challenge that we face. I think the hodgepodge of state regulations is where we're gonna see the largest level of difficulty, the privacy concerns. But from a theoretical standpoint, being able to get at your own data, what the mechanism is, what the mechanism looks like. I've talked with some groups that are using, you know, like Snowflake clean rooms and things to be able to have that neutralized data set that's maybe stripped of PII and being broken down from a performance perspective.
So this is where I think it starts getting super nuanced into what do you actually back to Ian's original statements at the beginning. This is what you actually need? What would you say you do here, right? Like what do you actually need at the beginning of this? And then we can work our way through how to get there and I know you got that office face joke. Clearly that one planned well.
But that's the way that I start to look at these things, right? Like if you can start with where you actually want to go with the data, what do you want to understand? Guys like me don't know until we start playing with the data and we start moving things around. But often I tell people, like, don't give me anything that I'm not supposed to see. Like, I don't need that. I just need a couple of fields for me to understand how to stratify a large data set so that I can start to understand what's within it, and then I can start thinking about how I would visualize that data.
Data itself, which feels to me like data visualization might be a whole separate webinar that we need to be talking about for the second half of the year. But gentlemen, I really do appreciate you guys coming on with me today, sharing your insights. This has been a great conversation. I know I pulled a couple of nuggets and insights from this. I hope our audience did as well. But as we go into wrap-up here, guys, any final statements for our audience today?
Anything that we didn't cover that you wanted to make sure we said today?
Anshul (50:33)
Well, I would say AI is here to make the life of collectors easy. yeah, that's what AI is here.
Adam Parks (50:39)
So don't fight it and submit to the digital overlord. Got it. Any final statements?
Ian (50:45)
Well, I think we've covered good topics and you know, I want to thank you, Adam, for asking such good questions today. I think this is a very interesting webinar. I'm really happy to be invited to it. So thank you for that. And yeah, I would just say, you know, let's keep it up and ⁓ keep asking these questions because I think they're gonna keep coming up over the next coming years.
Adam Parks (51:07)
Agreed. I think this is the first of many. And Darrell, any final statements? I really appreciate it's nice to get to know you today. I appreciate you joining us.
Darryl Brown (51:14)
Yeah,
Absolutely, and picking back off what Ian was saying, you brought up some pretty great topics. and it was a great discussion. I appreciate your invitation. Yeah.
Adam Parks (51:24)
Awesome guys. Well, I really do appreciate your time and insights today. I appreciate you joining us. For those of you that are watching live, we appreciate your time and attention. We'll be reposting, we'll be posting the replay to this next Friday so you can share that with your colleagues as well on LinkedIn and YouTube. But thank you everybody for your time and attention today. Guests just stick around for one minute when we end, and we'll see y'all again soon. Bye everyone.
Anshul (51:49)
Thank you. Thank you, Adam. Thank you, everyone. Bye-bye.
