LPRC - CrimeScience – The Weekly Review – Episode 158 with Dr. Read Hayes, Tom Meehan & Tony D’Onofrio Ft. Matt White

Episode Date: August 8, 2023

On this week’s episode of CrimeScience, co-host Tom Meehan interviews industry expert Matt White, Multitude Insight. Listen in to stay updated on hot topics in the industry and more! The post CrimeScience – The Weekly Review – Episode 158 with Dr. Read Hayes, Tom Meehan & Tony D’Onofrio Ft. Matt White appeared first on Loss Prevention Research Council.

Hi, everyone, and welcome to Crime Science. In this podcast, we explore the science of crime and the practical application of this science for loss prevention and asset protection practitioners, as well as other professionals. Hello, everybody. Welcome to another episode of the Crime Science Podcast. I'm really excited to be joined by Matt White with Multitude Insights. Hey, Matt, thanks for joining the podcast. Hey, happy to be here. Good to see you.

So Matt, before we get started, I know you're a member of the LPRC and a lot of the members that you've spoken to probably have an understanding of what Multitude Insights does. Why don't you give me a little background on yourself personally and the company? Yeah, sure. Happy to. And yeah, we've been members of LPRC for, I think, just over a year now. So probably one of the newer members. And we're happy to see the organization growing and attended lots of the fun events and whatnot. So for personal background, professionally, I spent about nine years in the United States Navy. I was a pilot there. I flew signals intelligence on reconnaissance aircraft.

So, you know, many of the people in this industry are, you know, direct in on loss prevention or come from a law enforcement background. I actually come from more of an intelligence background. So I did that for about nine years. Decided to make the jump from active duty over into civilian life and attended grad school. While I was working on my MBA, I got really interested in the problem of really law enforcement technology. How do police officers, investigators, and crime analysts work together with software across jurisdictions? And the answer, you know,

kind of during my research while I was at MIT was not very well. They don't do it very well at all. And so the company Multitude Insights was founded to address that problem. We have a principal product in that industry called Bulletin that allows agencies to share data across jurisdictional boundaries. And then naturally, we started looking into what is a lot of the type of crime that they're responding to. Shoplifting, organized retail crime came into focus for us at that point. So that's where we started making the jump and starting to build our kind of initial products in that space. Awesome. Awesome. And I think you have a really interesting past and kudos to you to kind of jump in. I'm always interested when I talk to folks in startup land that make a decision to get into the retail space. Yeah, it's not the first target market you go after, is it?

target market you go after, is it? Yeah. Yeah. And I, I'm fortunate. I work with a lot of different startups, both within retail and without, and I'm always intrigued, uh, really, really smart folks get together and somehow hone to a place that, you know, I call home and have for 25 years. So, um, retail is not what people think. It's a lot more complex. There's a lot more going on. And, um, you know, so I'm always excited when we have really smart folks that come from outside of retail. And so it's exciting to have you on the podcast and in the industry. Thank you. Yeah, thank you. Thank you.

I remember the first LPRC event I went to, I ran into you maybe next to the bar. I'm not sure. It's possible. And you were actually one of the people who kind of helped show me the ropes a little bit. Who's who in this industry, if you will. But yeah, it's been great to learn and we're happy to be here. Yeah, it's exciting. And it's a fantastic industry. And I think in our space, you really do have to understand how to navigate some of the things that are going on.

And today, that's kind of what I wanted to talk about, especially you being newer into the space and kind of tackling not a new problem, right? This has been a problem, but tackling it maybe in a different way. So I have a few just kind of questions and then we'll kind of talk through it. But how has the shifting demographics changed the type of tools that AP and LP professionals are using today from your viewpoint? Yeah. So this is, this is really interesting. The, you know, the, the younger folks coming into retail loss prevention, asset protection grew up with an iPhone in their hand, right? These are kind of middle to younger stage millennials and even early Gen Z.

They're used to consuming data and lots of it, but not necessarily in spreadsheets and tables. And so one of the things that Multitude Insights has done, both in our law enforcement product for our police officer customers and in our upcoming product for loss prevention, is deploy a really rich data scene, but in an easily consumable format. Okay, so that sounds like a lot of marketing jargon. What does that actually mean? It means simple interfaces that can show an insight about how many alarms you're having per day and how many individual pieces of monitoring equipment you have running at any given time at the highest level, but then be able to dive in deep underneath that kind of initial dashboard look.

We found as we were moving into space that there was not a ton of platforms that are actually making data consumable in a kind of human-centered design way, if that makes sense. So that's been our approach. We think that for these younger people to come into the industry, learn it quickly, they need tools of the trade that are something that they're native in. So that needs to be digital first. It needs to be cloud deployable. And it needs to be graphic user interface based. And that's our approach to be digital first. It needs to be cloud deployable and it needs to be graphic graphic user interface based. And that's our approach to everything we build. Yeah, and I think it's interesting because I probably am not I wouldn't say that, but I'm not necessarily your average LP guy.

But I think there are a lot of us that, although when I started my first job, there were no, there really, I didn't have email. Like that's my, my context of it, you know? So in, in the, in the nineties, there certainly were cell phones and email, but when we were working, it was, was far out. I, I have been predominantly, even it, even since, been in roles that were more technology focused. So for me, what you said resonates a little differently, where I grew up with VBA and learning how to do data the old way before really even you would do it in Excel and driving it through crystal reports and really getting to the point where we are today. So I feel like I've seen that evolution. So talking to you, I have a whole bunch of questions that some other folks I know that

are in similar paths would have. Just AI is the buzz today. It has been for a few years, but I think generative AI has really bought it to the mainstream. but I think generative AI has really bought it to the mainstream. So in reference to AI and more advanced technologies, how do you think it's going to change the industry for good? How do you see us deploying them? And what is the implications for the LPAP folks overall, in your opinion?

Yeah, I like the distinction between AI and generative AI, you know, chat GPT being the most popular instance of that, that people are probably familiar with. So I have an intern at MIT, he's a rising sophomore to give you a sense of his age and experience level. And I said, hey, in passing, I made a comment about, hey, wouldn't it be great if we could summarize this police report and turn it into something that they could quickly send to another agency? And, you know, oh, by the way, we want to develop something like that for retail as well. And two weeks later, he had a fully functioning model, completely trained on law enforcement and retail data. And it was spinning out reports that would have taken a previous team, you know, weeks, weeks, weeks,

and months and months of development time. And he did it by himself in two weeks. So what does that mean for the industry? It means that we'll be able to deploy a lot more capability quickly if companies have the appropriate people on staff. Is it going to take your job tomorrow? No. There's a lot of conversation about future of work and what that looks like, and that's a different podcast. But I think what it does do is allow LP and AP vendors and the people that are in the field as well, actually frontline folks, to deploy a lot more capability. You know, facial recognition is a big one where these models are very adept at identifying people and linking them between

cases and being able to say, you know, there's a $500 shrink event here and a $1,000 shrink event there. I think being able to draw the connections between otherwise unconnected events will be one of the prime uses of AI. Where do we need to be careful? We need to be careful about how we use it ethically. It's definitely fraught with risk of pinning something on somebody that they didn't actually do, unless you still have the classic steps of any good LP professional that can observe the whole events and track it all the way in and all the way out, that stuff is still critical. So using enabling technologies to do that is important. AI will make everything faster. I think that's the main thing. Yeah. So when you talk about AI,

and I think this probably wasn't the planned event, and you're right, we should do another podcast on AI itself. If you check out the orange, that's AI at its best on the video. It's trying to figure out the lighting versus that. So I would love to do another podcast, but since we said it, I think it's important because it's a common thing that comes up. Generative AI specific to large language models is really just a huge model predicting what it thinks you want. So just a ton of language,

and it's just really, while it's remarkably interesting to all of us, it's a predictive model. Yeah, that's right. And maybe even if I can add a piece of nuance. From your viewpoint, and I, again, I love how when we're talking about this, the computer's on its own. From your viewpoint, do you find that some of your customers today or folks that you're talking to immediately go to what's in the news or the chat, the open AI or the orthopod of like, there's just so much generative AI out there and it's not new. It's just that because it became mainstream and the media picked it up. Do you find that that becomes more of a topic of conversation than it did six months ago? Yeah. So I think like AI fluency has picked up. People are hearing the words, they're learning how to use them. They're understanding what they mean. And so your

average customer can ask a question that six months ago, they didn't know they needed to ask, which is like, how are you using AI in this product? How are you connecting the dots for me in ways that my analysts who have 20 years of experience couldn't already do? So whether that's through processing data more efficiently and drawing conclusions more quickly, or whether it's some other capability,

we generally have to be able to answer that question now. The company that I started, Multitude Insights, we've only been around just around two years. So in the lifetime of the company, it's already changed twice. So that part is pretty wild to me. Yeah, so it's, I guess, interesting to see the evolution from the customer.

If I can add a wrinkle of nuance though, the real capability jump is that these generative AI models give the average human being who is non-technically fluent the capability to interact with a computer the way that programmers have for a long time, right? So you can write a plain English sentence to tell the computer what you wanted to do, whereas before you needed to write 1500 lines of code, that's very, that's the thing that is making all of this possible. Yeah. I, I, I, I,

I really think we should because I want to talk about more about multitude insights and what you're doing, but I really think we should have a, a just a podcast about AI and maybe it'd just be you. Maybe we can invite other people because it's a topic that I talk about quite a bit. And I actually, I think in the next couple of weeks, you'll see a few articles that I wrote just kind of, you know, talking about talking about it. I also think that it's important you mentioned model that, you know, at its at least and I'll say in my definition of AI as a computer replicating human behavior.

And machine learning is not necessarily the same thing. They go together. But when it often comes up and people intertwine them, I say a microwave, an ATM is technically AI for all intents and purposes. When you start to get into complex models, that's when you get to take advantage of a machine replicating human behavior by getting really good models. So that's often the conversation I have. And with generative AI, there is a lot of misconceptions about what it's actually doing. And this whole thought process of, while I think chat GPT is phenomenal, this whole thought process of,

while I think chat GPT is phenomenal, there's no set in nature at all. It does not possess, the models are not designed to replicate human behavior the way people think they are. They're just predicting. That's why you can trick it. That's why it can make mistakes. And so it's so interesting.

But like I said, we're gonna table a lot of that. So now that when we're talking about your piece of it, this kind of goes into the next thing. In the past couple of years, this has changed. But I would say prior to three years ago, there weren't a lot of young, innovative. And when I say young, I don't mean necessarily young founding, but certainly startup-driven, young founding, but certainly, you know, startup driven Silicon Valley, or depending on where you, you know, MIT backed or heart, you know, Harvard backed startups in the LP space.

You know, why? And we started with this. I mean, why? Why did you your team decide to tackle AP? You know, law enforcement, you could have probably done a lot of different things with your past. And I know some of the folks you work with. So what drove you to this? which is law enforcement, right? So I got into law enforcement, applying a lot of the lessons learned in the intelligence space that I picked up on DOD side of things. There's an intelligence cycle concept in the Department of Defense world where you plan your needs out, your directions, you go collect, you process that information, you exploit that information, and then you run another analysis, and then you disseminate it, and then you run another analysis and then you disseminate it and then you do it again. And that's the whole cycle. Right. We wanted to apply that to law enforcement in a more appropriate way. As we started to work with these agencies, just a massive amount of headache and paperwork and friction was we start to notice that when

they would interact with retail loss prevention, even though these are often, they know each other well, they sometimes are cut from the exact same cloth. They want to work together. They're frustrated about how they can work together. There's a big loss event and it generates a whole bunch of paperwork on the law enforcement side and on the LP side, it just looked like a lot of pain and something that needed to be optimized. And so we had one of our agencies that we work with here in Massachusetts, and I'm in the greater Boston area more generally, but one of these agencies said, hey, like, is there a way that you could help us get data out of these retailers so that they could hand off cases to us that we actually could go prosecute and we're not wasting time? Right. Don't have to spend four hours going down to the Home Depot and, you know, talking to the LP team and do all this.

Like, can we like how do we smooth that out? And so that's really what got us into LPAP and start kind of experimenting with some of the other big players in the space is how do we take what's in your case management system and distill it down to what the police actually need? We don't need all 140 pages of your report, most likely, right? And then how do we actually get that in front of a law enforcement professional and have them take it all the way up and actually get a prosecution out of it, right? Like, how do we actually discourage theft? That problem is super interesting to solve.

It's super fun. There's a lot of technically innovative ways to do that that we're deploying from our MIT background. That's what we're excited about doing. excited about doing. So I think just to, to, for you right now, being fairly new in the space, what are you working on right now that you're most excited about? Or do you think that the listeners would get something out of it? Or maybe I don't know how much you could share about some of the pilots or some of the customers you're working with and what problems you're helping them solve in the real world. Yeah. Yeah. So without naming any names, we're working with a couple of different other vendors. I think you may even be people that listen to this type of podcast to take the data that a

retailer does have, package it up nicely, securely, digitally send it to their local law enforcement agency, provide updates on the status of that case as it moves through the justice system, and then close the loop and send that back into the retailer, right? So we're working, we're an early stage company. The product is called Stoplift, S-T-O-P-L-I-F-T. And the whole goal is to facilitate that transfer of information out of the retailer securely into your local law enforcement. do you have this photo? Do you have this thing? Can we add this to the case? And do it in a way that is compliant with privacy standards and do it in a way that is also compliant with CJIS, CJIS standards for the law enforcement agency as well. So that's what I'm absolutely pumped about is deploying that. I think some of your listeners who that sounds interesting to will be a bit more of a public rollout of that kind of Q4, Q1 timeframe.

Awesome. And so what I mean, you there's a lot of different folks that listen to the podcast. Most of them are from the retail loss prevention or law enforcement space, but we do have some academic listeners and folks all over the world listening as someone that, you know, started the company as a founder, you know, what do you think the top, you know, the top entry to barrier has been for you? What's been the most challenging thing getting into this space? Yeah. How much time do you have? yeah um how much time do you have um no a couple of things i mean like uh the agency or the the folks that are that have been in lp and ap for a while are i think we've faced a lot of skepticism like oh who are you guys oh you're new oh you don't come from lp you don't come from retail uh so i got a lot of uh well yeah sure well

maybe we could talk to you but like you know it's a lot of that kind of um i would just say it wasn't open arms per se not that people were being uh hostile at all that's not the characterization it's just like skeptical all right what are you going to do that's going to help us out um the way that we have uh kind of tried to overcome that is we built, you know, an advisory board of industry professionals that are able to open some doors for us. Like, hey, how do I get the meeting that I need to, you know, to get to that unlock where you actually will listen to what we're saying and hopefully actually be able to work together. So that was definitely one of the problems. The other is retail is a massive industry.

It's sometimes difficult to know where to start. Vendors run some pretty tight margins. The retailers themselves, I should say, excuse me, run some pretty tight margins. And so there's not necessarily spend for trial products. And so we've had to navigate that, too, is how do we convince a retailer to actually roll out some innovative technology? Because it's not always a, it's sometimes just a cost center,

not necessarily a profit center for them. Yeah, I think we've got about a minute left. So one, I want to thank you for coming on and that this is kind of my last, give you the last 30 seconds to end. But you started to lean into this is, you know, your U.S. Navy career and your MIT background are probably counterintuitive, probably were some of those roadblocks. But what are, you know, what's a lesson that you learned in the U.S. Navy that is

applicable to the LPAP space? And we'll close with that one. Yeah. is applicable to the LPAP space. And we'll close with that one. Yeah. So the Navy, you know, a couple of like really powerful frameworks. Many of your listeners are probably familiar with the OODA loop, right?

And as an aviator, we're super tuned into the observe, orient, decide, act loop, right? I've tried to apply those loops to all of our product decisions that we make. So once we get through those initial, who are you and why are you in LP space questions, when people look at our products and they see that thoughtful intelligence angle, they tend to say, yeah, that's actually something I'm interested in. So I think being able to apply those, what are war fighting techniques to this industry has been kind of a nice plus of the military background, if you will. So definitely something like that. Awesome, yeah, that definitely resonates.

So Matt, thank you so much for being on. I met what I said, I really wanna have you back on to talk about AI and a little bit more about Multitude Insights. I appreciate you joining us. Thank you again for coming. Thanks, Tom. Appreciate it.

A lot of fun. And yeah, let's do the AI podcast. Thanks for listening to the Crime Science Podcast presented by the Loss Prevention Research Council. If you enjoyed today's episode, you can find more crime science episodes and valuable information at lpresearch.org. The content provided in the Crime Science Podcast is for informational purposes only and is not a substitute for legal,

financial, or other advice. Views expressed by guests of the Crime Science Podcast are those of the authors and do not reflect the opinions or positions of the Loss Prevention Research Council.