1
00:00:03.110 --> 00:00:14.490
Scott Heath: Hello, everyone, and welcome to our webinar today. Kyc, onboarding. And what we're gonna find out today is how machine learning and AI can help us do that
2
00:00:14.630 --> 00:00:31.499
Scott Heath: today. I've got with me Chris Lacava, who is going to help me navigate sort of a bigger process on Kyc. Investigation and the overall process of financial crime, prevention, detection and investigation and prosecution.
3
00:00:32.340 --> 00:00:49.090
Scott Heath: So without further ado, this is what we have teed up today. We're going to talk a little bit about sort of why we we need to care about this and and what that means in the world of Kyc. And then, more importantly, now, we're going to talk a little bit about things that teams
4
00:00:49.140 --> 00:01:09.470
Scott Heath: do today. And basically, how can we make that better? How can we make it easy. How can we bolt on to existing systems? And and then, quite frankly, and then we're gonna see a little bit of art of the possible. So Chris is gonna walk us through a day in the life in a new world. What? What things you could possibly do? With some of this new data.
5
00:01:09.880 --> 00:01:14.180
Scott Heath: So with that, what we'll do is we'll start here right? We've seen
6
00:01:14.180 --> 00:01:19.489
Scott Heath: this, and if you've joined us before for some of these, this screen hasn't changed much, and that is that this
7
00:01:19.490 --> 00:01:44.329
Scott Heath: entire ecosystem of whether it's fraud or credit card or cyber crypto anti money laundering right? They are all now, sort of convening. And this sort of centrality of Kyc and screening is really the 1st step. If we if we get that right, then we keep these people out and
8
00:01:44.330 --> 00:02:02.690
Scott Heath: and they can't do those things. But it's getting increasingly more difficult. And the bad guys are starting to realize sort of how some of these operations work, and they're able to sort of get around them or find loopholes. But again, the numbers are getting bigger, and it's getting a bit scarier when it comes to Kyc.
9
00:02:02.930 --> 00:02:18.139
Scott Heath: right? That's that process we're going to see here in a minute. But but now they've started to ratchet down. They've started to press in on. It's no longer the fine at the end of the prosecution. It's now you're not doing enough upfront
10
00:02:18.499 --> 00:02:38.270
Scott Heath: in your Kyc. Did you really do the enhanced due diligence? Did you? Do sip the proper way right? And so the point of this is again, the numbers are getting bigger and the processes are getting more complex. So they're dropping more legislation on us. You know, like the beneficial ownership.
11
00:02:38.430 --> 00:02:49.140
Scott Heath: Now, these are some of the elements that that most everybody understands is, how do we need to do these things? How do we find that beneficial ownership in the Cdd, how do we sort of do that?
12
00:02:49.449 --> 00:03:13.279
Scott Heath: Edd, or even bigger, is, how do we then sort of look inside of the firm and external of the firm and tie these things together, and and they're all sort of started in the cip. But again, if you're in the business, you know, you have to do these different elements right? These elements are core for what everyone does. Now, the way that you do these is what we're going to talk about today, and
13
00:03:13.280 --> 00:03:15.359
Scott Heath: why they matter so much.
14
00:03:15.658 --> 00:03:23.120
Scott Heath: In the new, in the new sort of digital world is, how can we enhance that? How can we make it easy for our teams.
15
00:03:23.350 --> 00:03:33.919
Scott Heath: and the reason why, again, it's sort of backstopping us right, which is now we're seeing something called Framl. And I just love this new term, which is that ability to take fraud
16
00:03:34.090 --> 00:04:02.230
Scott Heath: right and anti money laundering and quite frankly, cyber and some of those other elements is they're all dependent on how we approach Kyc. But if you look at the numbers, they're very, very compelling right. If you look at the fact that every new customer that's coming to you, and you know that either 1% or more. And that's what we know. So you know, some of the the other folks have said that it could be as high as 5% or 10% depending on what
17
00:04:02.230 --> 00:04:04.120
Scott Heath: line of business you're in.
18
00:04:04.210 --> 00:04:26.339
Scott Heath: or your firm is in right. When you look at that percentage, and you know that they're in there in addition to what that means. By what is this cost? Right? That means by not stopping them, and they get in the door. The numbers are staggering. They're huge of either losses, and then fines which we don't really have categorized here. But they're big, right? And so it's that 1, 2 punch.
19
00:04:26.340 --> 00:04:35.289
Scott Heath: The other thing, then, is 31% of those onboarding forms are considered false, faked, or some sort of
20
00:04:35.629 --> 00:05:01.459
Scott Heath: augmentation that is not legitimate in order to create or obtain whether it's a line of credit or a line of margin, or whatever that may be right? So 31%, that's a big number, right? And then only 30% of firms that are out there today are really service gonna spend money on that. Well, what we want to share with you today is a bit more about how to do that.
21
00:05:01.630 --> 00:05:23.340
Scott Heath: Now, what happens is when the Kyc. Is wrong. The next step is false positives in either an investigation or an alert process, and regardless of what tool you have today, whether it's a custom or a spreadsheet or activize, or any of these other products. And they're all those are all trademarked, by the way. But those products.
22
00:05:23.671 --> 00:05:43.230
Scott Heath: Are kind of related to what I would refer to as garbage in garbage out. If you're not doing an effective job in the Kyc process or in one of those individual steps. Then it's very difficult now, to start to trim out false positives and then false positives can start to really eat an investigation team's lunch
23
00:05:43.230 --> 00:06:00.080
Scott Heath: number one. The time that it takes to do this is very unproductive. The capability now of finding those later becomes even more difficult, because then become entrenched. And now we start to spend an inordinate amount of money following false paths or false positives.
24
00:06:00.381 --> 00:06:20.570
Scott Heath: To be able to do that. And so what we see now is this ability to use some of the things we're going to talk about today to increase the productivity. Allow Kyc. To become more accurate, which in turn then drives down false positives as we go forward. So what we're going to see then, is Kyc. Is not only the tip of the spear.
25
00:06:20.760 --> 00:06:35.080
Scott Heath: but it actually, if done properly, can supremely lower false positives, increase our accuracy rate, and cut these by very large numbers and the human cost, which is investigators or firms that are doing this for a living
26
00:06:35.354 --> 00:06:56.800
Scott Heath: is very difficult to go back and recoup those false positives the time and the effort to go do that right. And so really, what we're now saying then is Kyc, is more than just simple onboarding is actually now that conduit, that 1st step that we need to get in front of to be able to do this. And then when you look at some of the numbers down here. It's that human cost.
27
00:06:56.980 --> 00:07:02.300
Scott Heath: So now that we've kind of laid the groundwork, why do we want to go do all this, we'll go to our next section
28
00:07:02.740 --> 00:07:28.450
Scott Heath: right? So what is it that some of our teams or our groups are saying so in Acams? What we see, then is there are the beneficiary most everybody has started to understand that. We we talked a little bit that in a previous webinar, but the number. One thing, then, is in Kyc. And screening is stopping. The bad guys, the syndicates, the groups, the individuals, whatever they may be before they get in
29
00:07:28.450 --> 00:07:53.300
Scott Heath: the other element. Here is back to that formal comment which again, I just love that term right which is now we need to start thinking in ways that integrate that. So the cyber team, the Aml team, the fraud fraud team, the transaction monitoring team. All of those teams really need to start sharing right. And we hear this now. In the Atms world. Right? So what are those things mean? And and then we start to see that government
30
00:07:53.300 --> 00:08:15.940
Scott Heath: actions are increasing. And now we're starting to see what what is Gen. AI, and we'll talk a little bit in a second about that. Those are the key points. But again, it's it's coming back to what we're talking about today, which is Kyc. And the prevention before it even gets in. Now, when you look at some of the fraud folks, they are saying that it's important for multiple industries, right? Because fraud
31
00:08:16.150 --> 00:08:39.499
Scott Heath: is a much bigger kind of an activity. But it very much ties into the money laundering now, because what we're starting to see then is they're coming in. They're looking normal. They may be doing credit card kinds of things and then starting to launder money. So that's why this whole inner operation has become such an important capability is that they may look like one thing. We've had
32
00:08:39.500 --> 00:09:04.469
Scott Heath: a customer example where they looked normal. But when you run some of the math that we're gonna talk about today, and some of the things that Chris is gonna show us. You start to see patterns that were not normally there. And the reason why this framble approach instead of swivel cheering over to the person next to you in another group, you need to integrate the data because there are things they are doing now. That are there, and you can see it down here. Data gaps are too hard. There's too much
33
00:09:04.470 --> 00:09:25.870
Scott Heath: data. It's too complex. A human can no longer sort of rain all those things in in a spreadsheet. We have to be able to use the power we have to be able to use the system in the right way to be able to do that. So now, that's really sort of where we're going. And again, our groups that we're part of and if you're part of these groups, you'll start to hear those more.
34
00:09:26.420 --> 00:09:33.889
Scott Heath: One of the things that we can at Xpero empirically point out is that by simply seemingly simply
35
00:09:33.890 --> 00:09:58.180
Scott Heath: creating a better Kyc experience, creating a way to find that information run and use the appropriate AI and Ml. Upfront allows us to do that. And how do we do this now? So underneath the covers. What we start to see then, is, if I were just to use machine learning. I see a boost. Well, that's great. That means that it is working. Now, some of those can be bigger.
36
00:09:58.180 --> 00:10:22.949
Scott Heath: And again, this slide is meant to be. Your mileage may vary. But the most important thing, then, is when we start to see now this connected data. That is what graph databases do for us and graph algorithms in in addition to machine learning, start to show me the connections for whether, again, they're a new person or an existing, or or they've been in our firm for a long time. Those connections tell us
37
00:10:22.950 --> 00:10:33.320
Scott Heath: as much as potentially what we find in some of the pure machine learning upfront. But then the most important part here is the human. How does the human
38
00:10:33.530 --> 00:10:47.170
Scott Heath: communicate with data and those connections and start to give insights that now can give us an enormous boost in both this accuracy. But it all starts with Kyc.
39
00:10:47.440 --> 00:11:02.200
Scott Heath: Now, the second thing, then, is downstream from this. Yes, we started, and there are many steps in between. But this is empirical data to say that if you get your Kyc correct, if you use some of these things upfront
40
00:11:02.200 --> 00:11:23.779
Scott Heath: right now we can see in this case. And again, your mileage may vary, but this is an actual case. 83% false, positive reduction. That is enormous. How did we do that? That's back to combining those different elements and being able to connect that data seems very straightforward and very simplistic right. But it is game changing.
41
00:11:23.780 --> 00:11:46.920
Scott Heath: The other big sort of thing over on the right, then, is by using the appropriate amount of machine learning. And AI. We are not creating additional risk. That's the beauty of this is now using this and sort of spending a little bit more horsepower if you will upfront. And the other key here that Chris is going to show us is it doesn't increase complexity. That's another big thing.
42
00:11:47.219 --> 00:11:55.319
Scott Heath: That this does is it makes it simple and makes it easy. But the data and the coordination of the system is what gives us this huge
43
00:11:55.320 --> 00:12:20.060
Scott Heath: boost. So this is the goal. Right? This is the the Holy Grail. This is what we're all after, which is to drastically reduce false positives, increase the productivity, and then those of us that are in risk or or compliance, or or other kinds of model validation. Right? We need to make sure that there's no additional risk, because we have to go talk to you know whether it's Vincent or a regulator. Right? We need to be able to prove this, and that's the other sort of model.
44
00:12:20.970 --> 00:12:45.929
Scott Heath: So as we get into that, that's what we're going to talk about today. What does experience bring to the table? What is our investigation capability? But more importantly, today, we're going to focus on Kyc, and we're gonna walk through this. So these are all things that typically matter is, it's got to increase that accuracy. It's got to decrease those false positives. I've got to be able to bolt it on to what I have today, and so whether you have spreadsheets or custom systems. Or again, we talked about
45
00:12:45.930 --> 00:12:52.230
Scott Heath: those other vendors. Right? Those are all great products. But you need to bolt it on, because this is coming at you fast.
46
00:12:52.508 --> 00:13:03.099
Scott Heath: So what we've done now is we've built this in a modular way that you can buy only what you need and incorporate some of these things without sort of doing all of it, or ripping and replacement.
47
00:13:03.470 --> 00:13:18.180
Scott Heath: Now, why do we want to do this? Well, what you see over on the left now is what we're going to talk about today. We'll talk about Kyc. And Id and those kinds of things upfront. But what you see there down below is that all of these other dependent? And whether your firm has these as business units
48
00:13:18.180 --> 00:13:41.800
Scott Heath: or business functions right, they all need to be accomplished. And over on the right. Now, you see, those are the technologies that we now have at our fingertips. Right? We have this now incredible capability or horsepower. And what Xpero connected does is it? It connects those things in the middle. It's got to be simple. It's got to be end to end. It's got to have that horsepower
49
00:13:42.070 --> 00:14:05.840
Scott Heath: and it's got to give that user whether you're in the Kyc on boarding team or you're some hybrid, or you're in machine learning team, or you're part of the Aml or fraud, whatever they may be. You need to have a place where you can see and how you can coordinate to do your job. But, more importantly, the system needs to use those things over on the right. So that's really kind of that 1st step on this is, how are we going to get there
50
00:14:06.480 --> 00:14:24.550
Scott Heath: now? Most of us that are you know, familiar with what this process looks like there are several very specific steps, and as we've seen right, the government is mandating what those are, but a lot of folks will see the sip process as something very simplistic, and it should be. It should be fast
51
00:14:24.801 --> 00:14:48.730
Scott Heath: and it should be straightforward, right, and it should allow us to do that. But as you move to the right, what we start to see then, is a bit more due diligence. Now, when we start to connect those dots. We need to score what we're finding so that we can do some determination. Do I want to let this person in? Do I not want to let this person in. Do I want to let them in? Provisionally right? What are those things? And then I need to share? You know that I'm
52
00:14:48.730 --> 00:14:52.939
Scott Heath: transparent, and and I'm not introducing any sort of bias in this process.
53
00:14:52.940 --> 00:15:20.040
Scott Heath: But as I move to the right, things start to get a bit more complicated on how we perhaps enhance that. How do we go? Look at government lists. Well, one of the things we'll talk about today, Chris will show us is things like Dow Jones. Right? Their data, is, is available, and there are lots of other 3rd parties. Dow Jones has several products that we that we'll talk a little bit about to go. Do that. But again your mileage may vary. You may be subscribed to several people.
54
00:15:20.130 --> 00:15:34.439
Scott Heath: but the call here is that each of these different steps today many customers may have point solutions, or we'll see an example here in a minute where it's all on a spreadsheet. But as I get over to the right now, when you see the big sort of magenta bar.
55
00:15:34.870 --> 00:15:48.009
Scott Heath: there's typically a break, there's not a tie in. Those are all points, things that have happened. Somebody has run through a checkbox, and I've done that in a spreadsheet, right? But I have not tied that into an investigation. So it's extremely difficult
56
00:15:48.280 --> 00:15:57.939
Scott Heath: to go back and say, when and how did we let this person in? Was there something that we need to do? Can we enhance that? Can we be better? Can we predict? Can we
57
00:15:58.276 --> 00:16:19.129
Scott Heath: protect ourselves? If we found something? And that's really what this whole process is the secret sauce for what we can provide at Xpero is that capability to not only do these steps, these 1st 5 steps, but then allow, then, that tie in downstream for things that can be even more powerful. To save and lower those false positives.
58
00:16:19.370 --> 00:16:43.979
Scott Heath: Now, why, why is this so hard? Well, there's the poor onboarding team there in the middle. Look at! They're they're just. They're besieged right? They've got everything external and internal that they need to look at. So if it's a brand new customer man, I've got to go out and look at negative news. And Linkedin and I gotta go scrape that data. And if they filled online forms in, there are some things that we can do with Mac ids and things like that. But
59
00:16:43.980 --> 00:17:08.329
Scott Heath: that's a lot of data, right. That's really really heavy. And if we look to the right, then is the processing portion of that right? So if I'm doing continuous monitoring, or I'm starting to look at. If you know, a behavior changes, somebody needs a new credit. I've got to go do all those things over there. All of this takes time, and it takes money and those false positives. If I think I found something right, you can already see
60
00:17:08.329 --> 00:17:34.899
Scott Heath: why this is so incredibly cumbersome, right? And a lot of the party 3rd party folks down there. They offer great services, but their real problem is, they are a point in time. They are only giving you an individual lookup, perhaps, or they are saying at this moment, in time. You know, Scott Key. Looks fine, but tomorrow I'd have to do it again, right? And so on, and so forth. And so what we start to see then, is this starts to make the onboarding team or the Kyc. Team.
61
00:17:34.960 --> 00:17:39.620
Scott Heath: their life really difficult, right? And then you start to layer some of these things on.
62
00:17:39.660 --> 00:17:41.989
Scott Heath: So this is an example. One of our current.
63
00:17:42.100 --> 00:18:07.119
Scott Heath: Our previous customers was basically looking at how they could improve this. And for the 1st step of the 5, that may be 4 h. But as you get more into those, it could take as many as 48 h, and that sometimes is a good number, right? It can. It can take even longer days and weeks. In some cases, if the rabbit hole or the thread that I found in the Kyc. Starts to really become problematic.
64
00:18:07.390 --> 00:18:13.890
Scott Heath: But what we see, then is, I have all these individual elements, and what this customer was doing is they would go wrong. Equifax.
65
00:18:14.145 --> 00:18:38.370
Scott Heath: They would do a quick Google search and Google has some other advanced sip products. Now. They would have to then go look at those documents, and they would have to sort of eyeball them, and maybe they had some some abilities to sort of tag pictures and things like that to see what they are if they filled out forms online. They could go look at those Mac Ids, and then they would then use something like a Dow Jones, which is great. But it was all by hand.
66
00:18:38.430 --> 00:18:51.160
Scott Heath: and someone now is is responsible for stitching those things together. Now, what you see is those different teams. So if I have aml or fraud, and I have cyber. And then maybe I have machine learning teams that are out there.
67
00:18:51.200 --> 00:19:16.770
Scott Heath: All of those teams would have to take that spreadsheet rejoin it. Right go do their own investigation. But it was not electronically connected. And again, that was that magenta arrow that I showed was all of these were perfectly acceptable elements, however, they were not necessarily coordinated. The other thing is what are called non obvious relationships, very difficult in a spreadsheet to look at connections.
68
00:19:16.770 --> 00:19:23.350
Scott Heath: previous connections, accounts, transfers following money. If there's an existing customer trail those kinds of things
69
00:19:23.350 --> 00:19:37.799
Scott Heath: very, very, very, very difficult. Now this check the box right and that they, you know, they they did fine in in some of their audits, but they were still seeing a very large, both false, positive, and they were having a very sort of large overall.
70
00:19:37.800 --> 00:19:56.380
Scott Heath: Well, now, how did we do those things? That's what we're going to talk about next? But hopefully, everybody can relate to this right? Because this is fairly common. And many of our customers today do things like this. And some of them, you may say, Wow, this is even advanced right? So the ability to use those things over on the left and then join those some folks are, are, you know.
71
00:19:56.607 --> 00:20:03.649
Scott Heath: doing things that are even lesser. But but they're trying to put their finger in the dike. They're doing their best right. And so that's really what we're trying to do
72
00:20:03.720 --> 00:20:05.559
Scott Heath: now when it comes to
73
00:20:05.570 --> 00:20:14.970
Scott Heath: Jen AI, as we've all seen the hype curve on. This is out there right? And and Chris will show us here in a minute what that looks like, but realistically
74
00:20:14.970 --> 00:20:38.360
Scott Heath: like Icarus. That's kind of our little analogy over here, right is that if you think that you can use Gen. AI for everything, and you get too close to the sun. We all know what happens right? The answer is, No, no, that's really not the case. There are still so many things that Gen. AI is is working on. But there are some things that are very good, and most importantly is, we do not ever want to run afoul of Vincent or the Regulators right?
75
00:20:38.360 --> 00:20:56.729
Scott Heath: They have some fairly clear things that they're doing and providing guidance on that you can use it for, and that you should not use it, for now they're not going to come out and ban you from doing certain things, but they're certainly going to frown on that. Nobody wants an audit. When when you do, when you don't need one. Right? So what are those things. Here's a quick checklist.
76
00:20:57.020 --> 00:21:11.230
Scott Heath: right? Which is simply what we want to do, then is show the user things. If there are obvious or hidden things. We just simply want to uncover those. We want to bring that to the forefront. So the user can make the determination.
77
00:21:11.270 --> 00:21:34.350
Scott Heath: So the ability to show connections that were perhaps hidden highlight things that were perhaps incongruous, or or there might be an anomaly just simply highlight them right? Show us what those scores may be of something normal and something abnormal simply show that human to say, Well, normally, it should look like this. And this one is is not doing that
78
00:21:34.350 --> 00:21:59.259
Scott Heath: right and that ability to then, perhaps even track where you are in the process. Some of these are complicated internal processes and external processes, like we saw in that one diagram. It may just simply be keeping a a checklist of things that you should have done, that you have not done right. Simply show your journey to make sure that you are in compliance. Right? So those are all very good reasons and things that J. Jen a. In fact.
79
00:21:59.260 --> 00:22:11.529
Scott Heath: the word gen AI is probably overblown in those kinds of things. Those are things that most sort of prudent processes do today to have a checklist. They just may be not automated. Now, what we don't want to do
80
00:22:11.630 --> 00:22:36.719
Scott Heath: to start to reason for the human. We don't want to then start to make those determinations we don't ever want to, you know, be sitting with a somebody from Vincent and saying, well, the system decided, that's never a good word right. What we want to do then is is what that is. So again, that's what we don't want to do is we don't want to box in the user. We don't want to predetermine things that leads to bad bad paths if you will.
81
00:22:36.720 --> 00:22:51.650
Scott Heath: So now that we know kind of what those rules of the road are. Let's talk a little bit about sort of what we're going to do. So, Chris. What I'll do is I'll hand this over to you. And why don't you give us a day in the life? And then, when we're done, you and I can then unpack what what we're sort of checking out.
82
00:22:52.120 --> 00:22:56.400
Chris LaCava: Sounds great. Thanks, Scott. Let me share my screen here.
83
00:22:57.020 --> 00:23:11.589
Chris LaCava: and I'm gonna walk through a concept. That takes us through the Kyc process. And then we're actually gonna get into a little bit of the investigatory aspects of the next steps. And that's kind of
84
00:23:12.027 --> 00:23:32.062
Chris LaCava: where scott was mentioning that this highly connected data and bringing in another context and weaving, not just K. Yc. In the beginning of the process. But this kind of ongoing monitoring. So the used case I want to set up here is operations person,
85
00:23:32.906 --> 00:23:55.509
Chris LaCava: working in a compliance group and a bank. And I've come in looked at my daily triage queue. And I've seen that someone that I have on continuous monitoring. Ha! Basically something tripped there. They they? They failed a screen. And so I've clicked in now. And I'm looking at that
86
00:23:55.888 --> 00:24:18.209
Chris LaCava: that record of that customer who? Who? Who has just recently tripped that screen and I and just to orient everyone on what we're looking at here. I see the customer information up here. I see when the last screen was and then I see some other metadata and some some alerts around, basically like
87
00:24:18.210 --> 00:24:46.980
Chris LaCava: some of the details of what's happened here. I've gone through and clicked on the screening tab just to see the information around the the actual screens that were run. I see that continuous monitoring has been turned on for this individual and that there is a standard Edd, so that's an enhanced due diligence screen that runs in the background, basically as information comes in.
88
00:24:47.382 --> 00:25:16.809
Chris LaCava: On this particular customer. And I see, in fact, yes, that there was some trigger around Associated Risk that called to attention this particular instance, and this was an automated run, because we had that continuous monitoring turned on. And then I can kind of look through the history here and see that over the course of the years there have been other screens that have run and and and passed. So
89
00:25:17.176 --> 00:25:43.600
Chris LaCava: here's an example of one that was manually run, manually run when a mortgage application was put in. Through this, through this customer. There is also an automated run when there was a credit increase, and then, when the customer is onboarded way back in 2,018, they also passed that, and there are different levels of screening here that this system would hold as templates, plus
90
00:25:43.600 --> 00:26:10.490
Chris LaCava: the ability to create ad hoc, screening mechanisms where you pick different criteria to screen against. And so we've got just regular person identification here, or a a sip screen for the onboarding process that goes up to customer due diligence, and then the enhanced due diligence. As things ratchet up here. I'm just gonna click back on the history and just see what happened. And I can see here my data sources are selected
91
00:26:10.490 --> 00:26:32.110
Chris LaCava: in this case, and I see there were a few attributes that we matched. We used as matching patterns for that particular customer. And we we checked some open source records and left lexisnexis. And we and and we, we basically passed through that criteria and identified
92
00:26:32.663 --> 00:26:40.406
Chris LaCava: this this customer successfully, and passed that criteria for that level of screening.
93
00:26:41.150 --> 00:27:06.070
Chris LaCava: But if I go back here today, when I've come in and and started triaging some of the details of this. I see a few things. Not only that associated risk here with the severity of high. But there are a few other artifacts, and that this screening has produced that are brought to my attention. 1st of all, there's been a recent activity, where there's high frequency, training
94
00:27:06.070 --> 00:27:16.920
Chris LaCava: transactions, and an account that that this looked at, and then also on a credit report, there's a discrepancy. An address. So there's there's now a little bit
95
00:27:17.275 --> 00:27:29.350
Chris LaCava: of information hanging off of this that I can start to dig into, and details. And historically, I can look at that stuff, too. You can see here there are certain artifacts that are
96
00:27:29.360 --> 00:27:38.970
Chris LaCava: are connected to each of those screening runs for that Kyc pass. And I can go to my artifact vault and see the list of that stuff.
97
00:27:39.456 --> 00:27:53.260
Chris LaCava: Over the, you know, over the course of building that vault up through these screening exercises. But today I'm gonna start just looking at the the credit report and see what I can see here, as far as the discrepancy that's been called out.
98
00:27:53.590 --> 00:27:54.720
Chris LaCava: and
99
00:27:55.422 --> 00:27:57.987
Chris LaCava: what I see is that
100
00:27:58.890 --> 00:28:26.815
Chris LaCava: Their total open balance looks pretty high. And then also there was a mismatch here, or or inability to match a previous address on a credit report, and that hadn't tripped earlier. So I'm going to make a note of that might be nothing. You know, these credit reports get false data. From various systems every once in a while. And it's just something that that that gets out of sync. But in this case I'm just gonna keep an eye on it.
101
00:28:27.970 --> 00:28:54.310
Chris LaCava: And now I can kind of dig in a little bit more. But as Scott mentioned, we can pull in some relatively novel technology to help us summarize a lot of this information. Not necessarily guide us in the in in our inquiry, but definitely cut to the chase. On gathering a lot of this information, summarize it and and have a look at it. So I'm going to click on that.
102
00:28:54.400 --> 00:28:57.890
Chris LaCava: And it summarizes each of these issues.
103
00:28:58.542 --> 00:29:12.970
Chris LaCava: And says, this, this digital assistant, basically that I've launched over here has has given us. A breakdown of you know what it, what it could mean for high frequency transactions. And
104
00:29:13.040 --> 00:29:26.270
Chris LaCava: these these kind of like sanctioned connections again like this might be a few degrees out something for us to potentially dig into. But that mismatch address kind of
105
00:29:26.695 --> 00:29:34.769
Chris LaCava: clued me into a few things here. So these things together have some correlation that follow a pattern
106
00:29:34.770 --> 00:29:58.389
Chris LaCava: that warrants, maybe digging into this or escalating this in some way. So the system has also said the possible next steps, not a recommendation, but the possible next steps after this are that you could manually run another screen, but add more criteria, add more data, maybe tighten up some of the things that the the standard screening did.
107
00:29:58.390 --> 00:30:03.479
Chris LaCava: and just see if you can pull back a little bit more information. I can also just
108
00:30:03.650 --> 00:30:07.370
Chris LaCava: escalate this to compliance. Maybe I don't have enough
109
00:30:07.797 --> 00:30:27.252
Chris LaCava: information to as to escalate it to the, to to a case yet. And and have it basically be put in the queue for a formal investigation. But I'm going to hold on that for a moment, and then I could just ratchet up the monitoring. So right now it's based on a standard
110
00:30:27.600 --> 00:30:38.130
Chris LaCava: enhanced due diligence. Maybe I want that as like a high degree due diligence, or something that would run more frequently even
111
00:30:38.473 --> 00:30:47.410
Chris LaCava: so so those are the types of things that I could potentially address. But for for now I'm going to actually just run a a
112
00:30:47.410 --> 00:31:06.309
Chris LaCava: something with a little bit more criteria. So now I'm in this this logic builder. Basically that allows me to. It's gonna walk me through a wizard to to to basically build the screening criteria for that Kyc pass. I already selected the customer and the level
113
00:31:06.658 --> 00:31:25.490
Chris LaCava: but I have certain fields on the customer, and this this is a subset. I could add more if I if I wanted to. And I can weight these differently. And what that's essentially gonna do is go through the subsequent data sources, internal and external. And it's going to attempt to match
114
00:31:25.955 --> 00:31:31.420
Chris LaCava: and do some entity resolution basically based on how I weight things.
115
00:31:31.782 --> 00:31:55.779
Chris LaCava: So I'm gonna you know, just keep this kind of middle of the road. And then I'm going to go to the next step, which is, select my my, my data sources. How? What am I? What paces am I gonna run this this customer through as far as what kind of criteria for for Kyc insight? So I can add some things
116
00:31:56.400 --> 00:31:57.500
Chris LaCava: and
117
00:31:58.680 --> 00:32:25.970
Chris LaCava: peps and sanctions corporate info potentially criminal records. We could add a bunch of them here. But I'm going to stick with this, and just just just just pull back a little bit more information than I did through the the standard automated version and if I even start to kick off this process. I already see there is a record that came back from from the pep scan, and it shows that
118
00:32:25.970 --> 00:32:46.260
Chris LaCava: this person there's a record actually, that shares a an address, and this may be an indication that that credit report flag that we saw wasn't a mistake or or incorrect record that got loaded in there. But maybe there is something to that I've seen now, too.
119
00:32:46.300 --> 00:32:57.840
Chris LaCava: 2 connections between addresses. So this is pretty interesting I could continue on with this scan. But now what I'm going to do is kick into an investigatory process.
120
00:32:57.950 --> 00:33:05.249
Chris LaCava: So I'm going to 1st look at. Let's look at these high frequency transactions. So I'm going to click on that.
121
00:33:05.400 --> 00:33:27.310
Chris LaCava: And I'm going to see the list of recent transactions. Now, this is just a warning, but I do see that there the account that we're looking at there has been a lot of activity over the course of the life of that account. And I can. I can even even narrow this down. And maybe let's go ahead and do that. I'm gonna look at high dollar transactions. Maybe maybe there's a set of
122
00:33:27.350 --> 00:33:53.699
Chris LaCava: patterns that are well known here that I'm going to. I'm going to see if this matches. So let's look at things that are maybe between 5,000 $9,000 here. And as I do that, I can see that yeah, in fact, that there are quite a bit of transactions there that meet that criteria, and I can even I can even ratchet this up or down, and see
123
00:33:53.700 --> 00:34:13.200
Chris LaCava: how the different aggregates here of money flowing from this account to another account work, and I can introspect these things and see some of the metadata around the list of transactions here and even filter this down to just a set of 2 accounts here, and just focus on this
124
00:34:13.199 --> 00:34:25.634
Chris LaCava: this dollar amount and see what the what the detail of each of those transactions are even drill into the the the the full detail of that transaction, if I want to. So
125
00:34:26.010 --> 00:34:50.249
Chris LaCava: you know we we want to focus on Kyc. But I guess I'm just saying here is that I can follow this now in an investigatory route, especially if this was escalated into something where it would go over to a compliance team that's geared up to do those type, those type of operational things. But let's go back to the Kyc stuff and look at the some of the other options we can dig into
126
00:34:50.610 --> 00:35:12.573
Chris LaCava: So there's a link to sanction entities again, you know that that might not be something that is is is nefarious, although now we're starting to build other information that may may link this person up to to a conclusion otherwise than than just kind of like the innocuous right? So
127
00:35:13.200 --> 00:35:23.339
Chris LaCava: I'm going to click through here, and I'm going to see how close these connections actually are. This is a high risk indication. So chances are it's going to be pretty close.
128
00:35:25.090 --> 00:35:27.380
Chris LaCava: And so, as I click over there.
129
00:35:27.930 --> 00:35:36.075
Chris LaCava: I see that. Yeah, there is. There is a a actually, a direct connection to something that I that I have to be
130
00:35:36.590 --> 00:35:54.769
Chris LaCava: that I have to be conscious of here. But there are other stuff. There's there's an individual on a pep list. There's also sanctioned country information in here. So I'm gonna expand this network out and see how those things connect. And so let me expand it this way
131
00:35:54.890 --> 00:36:14.930
Chris LaCava: and and bring in some more connections. And I do, in fact. See, yeah, there is. There. There is the transactions that we're looking at before, and that is connected to a sanctioned person here on a pep list. Sorry you know our Pep list, and if I pull it out even more I can see
132
00:36:14.990 --> 00:36:24.489
Chris LaCava: that there is a sanctioned country. This is getting a little bit busy here, so let's knock it down and just see where exactly those connections are. I'm going to select the Pep list
133
00:36:24.630 --> 00:36:31.110
Chris LaCava: and the country here. And what I see is the individual that tripped in that that Kyc screen
134
00:36:31.200 --> 00:36:47.259
Chris LaCava: actually shows that they're they have an account that's connected to other accounts with someone on Pep list, and then a couple more degrees out a sanctioned country. This is probably enough to to move this to the next step.
135
00:36:47.410 --> 00:36:51.160
Chris LaCava: And that's that's how this kind of
136
00:36:51.180 --> 00:37:05.240
Chris LaCava: this lens of Kyc, and also this other set of data information, the Framl kind of sets of data that Scott had mentioned are starting to merge. And we can start to see these workflows even
137
00:37:05.360 --> 00:37:18.400
Chris LaCava: even across multiple, different. you know, investigatory bodies or business units within a a set of compliance operation
138
00:37:18.757 --> 00:37:37.340
Chris LaCava: organizations within a bank, and just to kind of underscore that we could take this to another another degree, which is, instead of launching into that view, we could actually even look at this from a lens of of maybe some cyber incursion. Right? As Scott mentioned, a lot of stuff
139
00:37:37.460 --> 00:38:02.119
Chris LaCava: starts in one type of nefarious set of activities and quickly moves to others by design. So here I have some incursion information. I click on that. I can kind of see here the same kind of information that we were looking at before, with just the the Pep score and the the Aml. Sets of information that we overlaid on that network
140
00:38:02.551 --> 00:38:21.879
Chris LaCava: set of of data, and we see that there there was a set of incursions. If we look at the the the data across all of our different data sets to to look at, at at cyber, at cyber based activity, and we see that there is a known
141
00:38:21.950 --> 00:38:27.699
Chris LaCava: nefarious Mac Id that has been connected to to a ping on
142
00:38:28.116 --> 00:38:52.649
Chris LaCava: on on one of these Ips that we're watching. But if we pull in some of that Aml data now, we also can link it to things like an investigation and other sorts of things, like someone on a sanction list in this case, or a company that is on a pep list. So or connected someone on a pep list. So you can see how, as we layer this stuff over
143
00:38:53.114 --> 00:39:14.019
Chris LaCava: the course of not just onboarding, but also ongoing, monitoring that the connection of of this of this data and looking at it in this linked analysis view, we can start to see, not just individual issues, but really rings and and coordinated organized
144
00:39:14.020 --> 00:39:23.299
Chris LaCava: financial crime and other kind of other kinds of activity that you want to proactively find and and remediate
145
00:39:23.930 --> 00:39:30.650
Chris LaCava: and and get all of it right. See all of the degrees of it. Not just the the specific person and the specific incident.
146
00:39:32.900 --> 00:39:45.690
Chris LaCava: So I'll kick it back to Scott, and we'll talk through some of the underlying technology and and techniques that we use with our customers to to paint some of these pictures for them, and enable some of these workflows.
147
00:39:46.100 --> 00:39:50.409
Scott Heath: Thanks, Chris. I I think the the most important thing.
148
00:39:51.330 --> 00:40:15.040
Scott Heath: from sort of what? What we're what we're showing right is this ability to to see and sort of what those those different elements are. Right. So as we're seeing now is is, how did we do this? Right? So in our new flow. One of the things that that we want to talk about is if the system can help us do this.
149
00:40:15.230 --> 00:40:35.509
Scott Heath: that's what we want to do. And so what you see now is that data? How do we go out? And we do that? It's very straightforward, right, which is, if we can very quickly look at that data. And and it doesn't have to be complex data. It could be as simple as Google and Google offers a couple of simple searches, right? So as we start to connect that data early.
150
00:40:35.770 --> 00:40:57.409
Scott Heath: and as Chris showed us, we can start doing that matching. Now, what we do is we set ourselves up to start to enter that false, positive reduction. We've connected that data upfront. We've been able to sense it. We're able to in that magenta which is what Chris was showing you is at a simple level. I can see what things I've done, what things that I do to onboard. Them.
151
00:40:57.728 --> 00:41:15.540
Scott Heath: Where am I going? What that continuous is? Then it's a conduit in this sort of circular loop here is now I can risk and score and risk and score. And it now enhances those activities that are secondary. Now, it doesn't matter. So if you were to use, you know what we talked about today?
152
00:41:16.299 --> 00:41:20.260
Scott Heath: for that data connection. And then you're using one of those other systems.
153
00:41:20.400 --> 00:41:34.780
Scott Heath: The Xpero approach here is to work with those other systems. Right? You don't have to throw those out. You don't have to uncouple them. It's really making them aware and sharing that information so that downstream now we see that those investigations are much more enriched.
154
00:41:34.780 --> 00:41:57.130
Scott Heath: And we're able to show those connections, so that if we do go to sar, and we do need to go prosecute. It's all there, right. That entire lineage has been enhanced and now documented. And we've used Gen. A in the right way. So again, this is not a huge shift in what most people are doing today. It is simply an automated version of what that is.
155
00:41:57.560 --> 00:42:21.489
Scott Heath: Now, why does this matter, and one of the other things that connected from expurro provides is a much bigger capability. We know that the investigator we know that the onboarding team is is really asking for data that as it technicians that have to bring this data to those teams, it's very difficult, because I now have to look at high speed search. I gotta look at web search. I gotta look at the way Google
156
00:42:21.630 --> 00:42:34.160
Scott Heath: finds that data. I've got to use graph algorithms and graph databases to show connections. Those transfers that Chris showed us and highlighted are game changing. But how did I do that? Right? I have to stitch that data together.
157
00:42:34.330 --> 00:42:50.030
Scott Heath: Then I have regular questions. I have tableau and those kinds of things that I need to show Kpis. And where are those generic data decisions. And then I've got to start doing things in streaming. So in many of our customers, they have handshakes where they have to see this data in 30 ms
158
00:42:50.306 --> 00:43:15.460
Scott Heath: to be able to stop a wire transfer or stop a credit card or stop an onboarding in what we're talking about here in Kyc, then I've got to go back in time. So if I'm an existing customer, that may be an enormous amount of data, maybe corporate data, maybe personal data. It may be a combination of both. Right? I have to go do that. Then I need to in in some of the newer things that we're looking at today is, where am I in space and time?
159
00:43:16.160 --> 00:43:39.289
Scott Heath: Was I in a Mac Id, or was I in a cyber world where I was being forwarded by a nefarious country. I need to track all that data and all of these different elements now, or why? The investigator and one of our customers. The quote was, I would like to take all of my investigators and let them be investigators instead of hunters for data. Right? They have to go do all those things as we saw previously in a spreadsheet.
160
00:43:39.340 --> 00:43:52.360
Scott Heath: So this is why, again, sort of underlining what Chris was showing us is so incredibly powerful is, I'm stitching those things together in what looks like a very simplistic, and it is indeed an easy to use kind of a front end.
161
00:43:52.430 --> 00:44:18.829
Scott Heath: Now, how do I do those things that Chris showed us behind the scenes. So he showed that that logic builder that alert capability. Now, what you're able to do is simply by looking at the patterns and those links and being able to look at those different elements over time and space. I can take that Kyc, data. Right? I can take what looks like simple signals, tie them together, look at them over time and space, and that
162
00:44:18.830 --> 00:44:31.589
Scott Heath: combined with sort of that. You know what looks like a sip in an Edd kind of a front end can now start to become something so much more right. I can start to now cut the time, reduce the false positives
163
00:44:31.886 --> 00:44:55.329
Scott Heath: and make quite frankly, some of these folks that that are doing this make their lives a bit more simplified. And easier. And this is how this is how we do that sort of behind the screen. So what we see now is, there's a place for everyone. There is a place for executives. There is a place for stringers, there is a place for investigators, and even it and machine learning folks. Everyone has a hand
164
00:44:55.760 --> 00:45:02.189
Scott Heath: and a view into what we want to do. Now, there are other things that Kyc. Is used for. It's used for customer 360.
165
00:45:02.456 --> 00:45:26.930
Scott Heath: You know, because of the fact that we're starting to do some of these other things, we can start to share that data out of the same system. And now this is for some of the technical folks in the crowd. This is kind of how it's orchestrated. What you see over there on the left is the Gen. AI. Now, what you saw, Chris, show you was an appropriate use of Gen. AI. It's simply connecting into these different views of the data. So whether you're in the Kyc. Team.
166
00:45:26.930 --> 00:45:45.790
Scott Heath: your Aml or credit card or cyber, or whatever they are, that jetpack can be turned on or off, depending on your firm's policy for what that looks like. But ultimately, what it's doing is it's connecting the data. It's allowing workflow. So notifications and triggers and humans to be involved in the loop and bosses and managers.
167
00:45:45.790 --> 00:45:59.469
Scott Heath: Right being able to do those things is very, very much part of this whole process, and if you don't have that it can be sometimes problematic. Then you get into the analytics and some of the more technical things right where the it folks and stitching some of those things together.
168
00:45:59.750 --> 00:46:24.349
Scott Heath: But in the end Xpero offers modules, and what you've seen today is that there are many ways to connect to that data. And we offer those in different sort of versions. And whether you and enjoy those today in other software systems you can. You can certainly bolt in one or 2 of ours or you can go for the whole thing. Really, it's kind of up to how your firm wants to do that. Now, what do those things do?
169
00:46:24.350 --> 00:46:46.819
Scott Heath: Chris started talking about advanced capabilities, right? Being able to prosecute, identify things over time for money flow, and and how time and and different kinds of things occur. Right helps us predict and find in the future that geography we talked about that entity matching, being able to simply sort of point out where anomalies are help, you predict.
170
00:46:46.820 --> 00:46:57.509
Scott Heath: can stop, so that in the future they have to get even more sophisticated. And you're one step ahead. And that's really what behavioral trends, and that that radius look like. So again.
171
00:46:57.940 --> 00:47:22.830
Scott Heath: that's kind of how the new world can be done hopefully. What you've seen today is it's fairly straightforward. To go. Do those kinds of sort of elements in a larger process to again lower those false positives. So if you're interested, we have some packages today that we're certainly, that you can do that are fast. So if you're on the fence, or you're interested in these kinds of things again, we have very simplistic ways.
172
00:47:23.324 --> 00:47:30.199
Scott Heath: That we can go do those things, and they're fast, and they're rapid, and you don't have to throw out your existing system. It simply bolts on
173
00:47:30.683 --> 00:47:33.670
Scott Heath: last slide sort of bringing it home right?
174
00:47:33.890 --> 00:48:03.760
Scott Heath: The real goal here, as we heard from the different groups, is Framl. Fusion is a thing Kyc. Is really the linchpin of how this whole process gets started. And so a little bit more effort, a little more automation upfront pays an enormous dividend at the end. And ultimately the humans involved in Kyc. Are the differentiators, but giving them more sophisticated tools, giving them the ability to sort of share with others. Is the main message from today.
175
00:48:03.980 --> 00:48:09.270
Scott Heath: So with that we'll wrap it up and take any questions that may be out there. But thanks so much for joining
176
00:48:09.370 --> 00:48:10.439
Scott Heath: over to you. Lord.
177
00:48:13.200 --> 00:48:21.630
Laura Smith: Yeah, thank you. Guys so much. If you missed the beginning. If you have any questions, feel free to drop them in the QA. Box at the bottom of your screen.
178
00:48:21.984 --> 00:48:24.580
Laura Smith: We'll give it a minute or 2 here.
179
00:48:25.343 --> 00:48:37.890
Laura Smith: Otherwise just as a reminder. We did record today's session and we'll send out an email. Follow up later this week with a copy of the recording. It'll also be posted on our website and social media.
180
00:48:38.286 --> 00:48:44.940
Laura Smith: I'm not seeing any questions come through. But if you have any feel free to reply to the email and we'll get in touch with you.
181
00:48:44.990 --> 00:48:51.570
Laura Smith: Thanks again to our panelists today, Scott and Chris, and we look forward to the next session.
182
00:48:53.570 --> 00:48:54.490
Scott Heath: Thanks, everybody.
