Podcast 198: Charlie Delingpole of ComplyAdvantage

A travesty of today’s financial system is how easy it is to launder money. Despite billions of dollars in expense and tens of thousands of people attacking this problem banks stop around 1% of money from financial crime. Incremental change is not going to help here, we need a new approach. Our next guest on the Lend Academy Podcast is Charlie Delingpole, the CEO and founder of ComplyAdvantage. His company is taking a different approach to catching financial crime. He has created an algorithmic approach to data gathering and an API-based system for compliance that allows them to work with fintechs, banks or even non-financial companies. In this podcast you will learn:

• The visit to a remittance store that led to the founding of ComplyAdvantage.
• Why it is so hard to catch money laundering criminals.
• What ComplyAdvantage does exactly.
• What makes ComplyAdvantage different to others in the industry.
• How they are able to interface with both legacy banks and new fintech companies.
• Why it was so important to build everything from scratch in house.
• What data their API sends back to their customers.
• Why globalization has made money laundering so much more difficult to catch.
• The three countries they have operations in today.
• Why they had to become a global company from day one.
• Charlie’s thoughts on machine learning and how important it is in their business.
• Why the opportunity for ComplyAdvantage goes far beyond financial services.
• The scale they are at today as far as API load.
• Who is backing them and how they were able to raise from A-list VCs.
• Why Charlie is optimistic we can solve the money laundering problem.
• What is next for ComplyAdvantage.

This episode of the Lend Academy Podcast is sponsored by LendIt Fintech Europe 2019, Europe’s leading event for innovation in financial services. Download a PDF of the transcription of Podcast 198 – Charlie Delingpole. [expand title=”Click to Read Podcast Transcription (Full Text Version) Below”] PODCAST TRANSCRIPTION SESSION NO. 198  –  CHARLIE DELINGPOLE Welcome to the Lend Academy Podcast, Episode No. 198. This is your host, Peter Renton, Founder of Lend Academy and Co-Founder of the LendIt Fintech Conference. (music) Today’s show is sponsored by LendIt Fintech Europe 2019, Europe’s leading event for innovation in financial services. It’s coming up on the 26th and 27th of September in London at the Business Design Centre. We’ve recently opened registration as well as speaker applications. You can find out more by going to lendit.com/europe. Peter Renton: Today on the show, I’m delighted to welcome Charlie Delingpole, he is the CEO and Founder of ComplyAdvantage. Now, ComplyAdvantage are a fascinating company, they are in the anti-money laundering space, or really the compliance space and they have developed new technology which allows them to connect with pretty much any financial institution and help them really manage their compliance risk when it comes to anti-money laundering. So we talk about how they do that, how he was able to build new technology from scratch and how he’s able to manage really integrating with very archaic technology at banks, as well as with some of the newer fintech companies. We also talk about what’s wrong with compliance today and how we can really get to a point where we’re 10 times or even 50 times better at it than what we are today. It was a fascinating interview, I hope you enjoy the show. Welcome to the podcast, Charlie! Charlie Delingpole: Great to be here, Peter. Peter: Okay, so I like to get these things started by giving the listeners a little bit of background about yourself. You’ve had an interesting career, this is not your first foray in fintech so why don’t you give us a little bit of background about yourself. Charlie: Yes, so I started my first company when I was 16, my second company was in the peer to peer lending space so I started MarketInvoice back in 2009. Back then, it wasn’t really fintech, it was just kind of invoice financing, but then very soon it was this maelstrom of sexiness which became fintech (Peter laughs). MarketInvoice was kind of a peer to peer invoice finance platform lending money to small companies, lent a few billion now, raised $35 million dollars from Santander and Barclays back in February. I started this company ComplyAdvantage about five years ago after MarketInvoice and we’re roughly 200 people and 450 clients and four offices around the world now. Peter: Okay, so we had Anil on the show last year so I think the listeners are pretty well aware of what MarketInvoice is all about, I guess the question is why did you leave MarketInvoice which obviously has grown a lot since you left so what was the impetus to leave and deciding to start ComplyAdvantage? Charlie: I think things were going very well at MarketInvoice, lots of opportunity elsewhere. Anil was doing a great job so I thought it best at that time to part ways and move on to something new and exciting. Peter: Okay, so then what was the ‘a-ha’ moment or the idea that led to the founding of ComplyAdvantage, what was, you know, the germination of the company? Charlie: So kind of the damascene moment, I think, was when I walked into a Somali remittance store in Mayfair in 2014. This company was based over at the Saudi Embassy, they had these big armed guards juxtaposed with these large trucks of cash moving money from from Somali and Afghan communities around London and just seeing the people there having to deal with processing huge amounts of payments which could be going to Al-Shabab, the Taliban, ISIS or could be going to your uncle and to see how terrible the software was and the data was and seeing how…for me, it was extremely obvious how terrible, also how consequential what they were doing was. Therefore, at that point, I had no real choice other than…like I’m just going to have to go and build this business because I think given my experience in the past, given what I knew at MarketInvoice in regards to how difficult it was to onboard investors, how difficult it was to understand the risk of which companies to lend to, given what I knew about the market at JP Morgan, given I knew about the potential in terms of technology and data from my first company and from working on MarketInvoice, I saw an amazing opportunity to restructure and reinvent something which is obviously being done in a really archaic and terrible way. Peter: So with that…it’s not only that, I mean one of the things about financial crime, I continue seeing this stat sort of bandied around, is that like we catch 1% or thereabouts of all of money laundering globally and we have these massive compliance organizations inside banks that don’t seem to do a very good job. So I guess maybe…before we get into ComplyAdvantage, why is it so hard, why is it so hard to catch money launderers? Charlie: So to an extent, it’s an arms race and it transcends every geography, every industry, every client type and it’s the full spectrum of human behavior so it could be Trump who today cut down on Iranian sanctions with oil, the next day it could be human trafficking, it could be wildlife trafficking, it could be trying to defraud a peer to peer lending exchange, it could be money laundering. And because money is so fungible and can be disguised and transmitted in so many different things and in different services, the nature of capital means it can morph and flow and therefore, it’s impossible to define and therefore it’s everywhere, as in you can disguise it in inflated invoices, you can do it with loans and, therefore, what you have is this impossible arms race between criminals, terrorists, money launderers and institutions who are the vectors and vehicles that they try to exploit. Peter: Okay, so then maybe we can segue into ComplyAdvantage and maybe you can tell us exactly what your company does. Charlie: So principally, what ComplyAdvantage does is help companies manage the risk around sanctions and money laundering. So on a very basic level it’s, am I allowed to lend money to this company? No, because they’re an Iranian general who’s funding the Iranian terrorist group, or no, because they’re the Venezuelan central government ambassador and therefore they’re prohibited from any kind of funds flowing to them. So, on the most basic level it’s, am I allowed to do business with this person? Peter: So what you do then is you have…I imagine you just have a massive database, right, that’s constantly being updated, I imagine. I mean, so are you really just focused on the binary decision on is this person a good and reputable person so I guess that’s one part of the question. Two, is then…what is the secret sauce? I imagine a lot of people have databases of bad actors so what is your secret sauce? Charlie: So the reason why everyone has failed in the past is because it’s a very, very challenging problem. So what you have now is hundreds of millions of dollars spent on manual labor and despite all of this expense, despite all this investment, despite all this technology being deployed, people still get fined billions of dollars. The terrorists and money launderers still win, right. So for whatever reason in the past, no one has managed to solve this problem and so to me, it was trying to build a company that was capable of succeeding where everyone else has failed, right, and so that was principally looking at the industry, understanding who is doing what,  the way they were doing it, the different incentives to each party, the different structural legacy systems they’re stuck with and trying to invent a solution and that’s what we’ve built over the past five years. So in summary, what we’ve done is what was seemingly impossible at the start which is build the entire system end-to-end which is the case management, the search algorithm, the API, the transaction monitoring, the payment analytics, the [inaudible] resolution and database. All of these things in a combined holistic system, we think, are now able to make a real dent in the challenge which has never been possible beforehand Peter: So not only that, you’ve got to not just build it, you’ve got to interface with existing systems which, I imagine are all very different. I mean, obviously, you’ve got some of the core banking processes, there’ll be some similarities, but how are you able to interface with everyone from…you know, you’ve got a legacy bank running COBOL to a fintech company that’s five years old or less and running some of the latest Python technology, how do you manage both? Charlie: So I think what people don’t often understand is that an API is the most important interface of all like while it’s very easy to build an astounding human interface in terms of the kind of web interface, in terms of catering to developers and the industrial process of managing millions of searches and the ability to get very high performance results from the API, being able to design that and have all the functionality around that, I think that’s a huge differentiator in terms of the way it’s constructed and the way it’s built, the way it’s maintained, the way it’s interfaced. And so we’ve invested huge amounts of resources into building what we think is the best API in the world and so all the clients we have integrate with the API. Peter: So it doesn’t matter whether they’re running COBOL or whether they’re running Python, they will integrate…they will be able to get the same results from interfacing with you, regardless? Charlie: Exactly, yes, exactly. Peter: Okay, okay. So I’m curious then…you know, before you guys came along, what sort of compliance…I mean, banks…as you said they’ve invested billions of dollars into this, what have they been doing and why is it so inadequate? Charlie: So a key dimension of what we’re doing is around all the data so what you have now is or in the time before we arrived was you had three or four main data providers and these data providers what they do is they really pride themselves on having 500 analysts and all these analysts kind of came from editorial backgrounds and they try and track…of the 7 billion people, they try and track and then maintain a database of say 5 million people they thought were high risk, right, so sanctions, political exposure, adverse media. So what that means is if you have 500 analysts and they spend a few minutes every four years, they can spend therefore 10 minutes on each person every four years which means that the breadth and depth of the data is completely not very helpful. Peter: (laughs) Right. Charlie: Whereas for me, starting off from scratch, I felt the only way to solve that was to do it via an algorithm. Of course, when you start building these algorithms, at the start they aren’t particularly effective, you have to invest huge amounts of resources and time to get them up to the stage where they’re better and more effective than the human researchers. If you can do that then that’s a huge inflection point and there are all sorts of other things you can do once that foundation infrastructure is built. So and then in terms of once you have that data then you can have the fusion between the data and the API and the search algorithm and the case management and so, architecturally, what you have is a step change in terms of what’s possible. So by rebuilding everything from scratch in-house in this kind of holistic system, we think we’re able to get much better results. Peter: Okay, so I want to actually just…if you could just talk us through an example how it works in practice. So someone comes along to a bank, for example, they’ve got $100,000 that they want to deposit or they want to invest or whatever, the bank says, okay, we need to do our KYC and AML type test. So this person enters in a whole bunch of data, then they send it to you through the API, what is it that you do exactly and what do you send bank to them? Charlie: So what we send back is profiles of people that we think match the entity that they’re trying to onboard and details of that person. So it could be Peter Renton, he’s involved in human trafficking, he launched an attack on Al Qaeda in (inaudible), all of these things about this person and they’ll say, okay, actually, different date of birth, different Peter Renton and therefore we’re going to remediate that profile as not being the correct match. The next match could be the correct Peter Renton in which case probably you don’t want to onboard him because he is a super high risk. Peter: Right, okay. So is the output from your API really just a binary yes/no decision, or do you send back like a much more rich kind of dataset and then the people at the other side make the yes/no decision? Charlie: What we send back is an entity profile so it could be a person or a company and they have to decide is that person or company the same as the one that they’re trying to onboard and they’re confronted with currently. So we don’t make the choices, it’s up to the lending company themselves to decide whether or not they want this person or company as a client. Peter: Right, right, okay. Charlie: The challenge becomes…if you’re on-boarding a million clients every year and if you already have a million clients that are already clients…because at any given point, a few of those could become terrorists, they could become money launderers, they could become subject to things like cuckoo smurfing, they could be taken over by criminal gangs, so you need to know when you’re maintaining your client base which of them become high risk… Peter: Right. Charlie: …because that could land you in jail and at the same time when you’re trying to grow the business and make money, you have this hidden risk which you’re not aware of. So it isn’t just the entity base risk profile, it’s also the kind of the transactional behavioral analysis. So let’s say you have two different companies who are transacting with each other as counterparts via a system then you have to be able to understand that risk as well or if they start sending money to Iran, or they start transacting in Iraqi Dinars so there’s both the entity risk and then there’s the ongoing monitoring risk and then there’s the behavioral risk as well. So we’re trying to help companies understand all those different attack vectors. Peter: Right. So I imagine you’ve got to be like Google in some ways, you’ve got to go out and just search the Internet for all these kinds of…I imagine you’ve got some pretty extensive different data sources. Someone may not be a money launderer today, they might be a completely law abiding citizen, they wouldn’t raise any red flags, and something happens and they get recruited by a terrorist organization or what have you and suddenly, they become one. So how are you building…are you just going out…I imagine it’s all automated, but you’re just going out, your services are going out scouring the globe for new cases of people who’ve become bad actors? Charlie: What you have absolutely is this trend whereby there’s this explosion of information and it could be in Swahili, it could be in Afghan, it could be in Russian and then you have…it could be Vladimir Putine in French, it could be Vladimir Putin in Cyrillic so you have all of these different people, all of these different sources, billions upon billions of new webpages, social media feeds, all of these different risks that suddenly appear and you have clients supplying to you from all over the world, from shell companies and so if you only have a team of say 50 people, how do you manage this enormous risk, right. So what we’re trying to do is make this as efficient as possible. I think it’s going to be difficult to completely eliminate the risk, but the more data sources and the more ability to make simple what is inherently complex then the easier and lower risk it will be for different lending companies and the more likely it is that they can stay in business. Peter: Sure, so I mean, there’s been a lot of talk this year…I’ve read many articles about like the trouble, I think it was in Estonia with the money laundering, I’m curious to get…what region, what parts of the world have the worst or have the most financial crime? Charlie: So we have clients now in 45 countries and what’s fascinating to me like traveling around is how each different jurisdiction is completely different in terms of the risks exposed. So in our Singapore office, for instance, a big risk is the Chinese abscond, the Skynet risk, those people who have been tracked by the Chinese government and then go to Indonesia or Malaysia, so there’s a risk there. In the Baltics we have a team focused on Latvia, Lithuania, Estonia and so the risk there is obviously all the Russian money coming in and then in South America it’s going to be different forms of drug gangs and so those aren’t necessarily siloed. So given the intensification of globalization, it could be the Mexican drug gang is working in league with a Russian money launderer who is then linked to a Chinese drug cartel… Peter: Right. Charlie: So I think as crime becomes more transnational and becomes more complex, the risks and consequences of criminal finances become all the more extreme. So it’s everywhere, it’s hyper local, but also global at the same time. Peter: Right, right. So what geographies do you operate in? I know in talking to you today you’re in New York, but you’re obviously not American, I know you’re British, but what geographies do you actually operate in today? Charlie: So we have three global hubs so New York, London and Singapore and the teams are kind of based out of there. Peter: Okay, okay. You said you’ve got clients in 45 countries so I guess whatever country is closest that’s where…you’re managing everything out of those three offices. Charlie: In terms of offering the best support and also in terms of understanding the local risks and also the local name matching requirements, the kind of languages, the data sources, each different regulation and regulator has different nuances which we have to understand and to operate in one country, you have to operate in the whole world because each local country will often need screening for global requirements. So it’s not good enough having local coverage, you have to be completely global from day one. Peter: Wow! That’s a decent sized barrier to entry, I imagine, for a lot of people, a lot of companies. Charlie: I think the barriers to entry in this business are I think huge and only getting larger in terms of reputation, in terms of adoption, in terms of data coverage, in terms of efficiency so I think clients are becoming more and more demanding as the risks and consequences become tougher and tougher. That was partly why I raised this $30 million dollars from Index in December is because right now, we’re roughly 200 people, but I think we want to take our tech team and development team…like try and double that this year, if we can. This year there’s a range of things we need to do and a sheer number of breaches and data sources and efficiency and effectiveness to the algorithms that clients demand is just only going to increase as time goes on. Peter: Right, right. We haven’t talked about it yet, but I imagine this is a big part of your technology and that is artificial intelligence, machine learning. You’re dealing with massive data sets so how are you using AI/ML in your business to help companies with all of this monitoring? Charlie: So for me, the whole AI thing is kind of very overhyped, but I think if you really understand what it actually does in production and how you can actually use it, it is actually fairly simple. So I think in ten years, people won’t really talk about machine learning as a kind of amazing thing, it will just be something that everyone does in the same way that SQL was amazing 20 years ago, but now it’s something that feeds everything. So I think in terms of how we actually use it. I guess there are different dimensions of machine learning in terms of supervised/unsupervised, natural language processing, entity resolution, anomaly detection like…all these things that are kind of tactical things which can help improve certain sub-sections of the platform, but to the extent…what we’ve basically built is this very broad system and you have to look at each different component of the system to understand in that specific juncture…can that be improved by a machine learning algorithm.  I think what we’ve done is use it in certain components, and I think probably had we not used that technology throughout our system, it probably wouldn’t work at all, right. So I think the most basic one is a classifier so what we have is a database of like 5 million people and a key challenge is….is this person that we just isolated the same as an existing person, so you have to look through when you have a new name, is this person the same as another one of these 5 million entities, right, so that’s a binary classifier and the 22:19 more training we feed that, the more accurate that is, right. So I can go through like entity resolution so if we have 5 million clients from a client then which of those existing clients is the same as the current ones. So all those things can be improved with technology and as Google, Facebook, Amazon release more and more algorithms, we think that our technology will only improve with time. Peter: Right, right, okay. So who is your target customer? Is your target customer every financial institution on the planet, or is it large banks, I mean you’ve got…I imagine it’s a big problem for just about everybody, but who are you specifically targeting? Charlie: So right now, we have like 400, 450 clients directly, but then part of them via reseller partners so lots of our clients will work through say like…for instance, Jumio announced they’re working with us so you can access our data via Jumio, right. So yeah, I think we work with lots of lending companies, payment companies, banking companies, but then also high risk corporates too, right. I think, ultimately this will be relevant to all companies so one of our earliest clients was Fedex in Zimbabwe who had to understand if they were dealing with the Mugabe regime so I think companies all over the world are becoming more and more concerned with the integrity and fidelity of their counterparties and as the risks increase, we think all companies will care very deeply about this. Peter: Right, so you are going beyond financial services. Obviously, you have Fedex, they’re not really a financial services firm, but that’s really interesting. That is a much bigger opportunity then. Charlie: Exactly and I think, to an extent all of these categories like credit, fraud, identity. In a way those distinctions are relatively artificial in terms of fraud. Perhaps 20% of credit losses are because of fraud or perhaps lots of losses from fraud are because of identity issues. So I think, over time, all these markets will merge into one rather than being distinct categories. Peter: Right, right, fair enough. So then can you give us the names of any of the clients you’re working with, I mean, are any of them publicly declaring that they’re partnering with you? Charlie: I think on our website…I think we’ve got Santander and Visa/Earthport and Holvi, and I think we’ve got LendInvest on there so yeah, kind of a very broad range of banking, payments, fintech companies. Peter: Right, right, okay. So then you’ve given us a sense of the number of clients. I’d be curious about…because some of these clients might have…you know, Santander obviously, millions and millions of clients, what is sort of the volume of inquiries that you get on a daily basis? Can you give us some sense of the scale that you’re dealing with right now? Charlie:  I think it’s roughly 100 a second, in terms of like API load. Peter: A hundred a second, okay, so what’s that…that’s 360,000 an hour then? Charlie: Yeah, but I think also it’s the numbers again who are monitoring daily, right, so a lot of clients are very concerned about the ongoing risk and have daily alerts set up as well. Peter: Right. Charlie: So that’s kind a decent milli-number of search every night. Peter: Right, okay. When I was doing some background on you guys and I see the name…is it Mimiro, is that how you say it, M-i-m-i-r-o? Charlie: Yeah. Peter: Is that a new brand or so what’s going on with ComplyAdvantage and Mimiro? Charlie: We’re in the process of sorting out all of the trademarks so there was a slight hiccup with the trademark, with a small Norwegian farming company (Peter laughs), who might have registered the same trademark at the same time, a week before us, so yeah, we’re trying to sort that out.(cross talking) Peter: The intention is to do a rebrand down the track then? Charlie: Yeah, I think so, yeah. Peter: Right, right. Okay, you mentioned Index Ventures and your Series B last year, I think it was just late last year, so I’m curious about the venture community, I mean, and obviously you’ve got some A-list backers. What do they see, tell us about that process and how you were able to land some of these top name VCs. Charlie: So we raised very early on from Balderton, I think then we were kind of 15 people and we had Tim Bunting and Suranga came on board at a very early stage and they invested $7 million, I think it was like three years ago and that was really instrumental in terms of bringing on the right people who really understood both the technology and the market opportunity and they really encouraged us to invest extremely heavily in the platform and technology. We didn’t really kind of take the product to market for a long, long time until we felt it was of sufficient quality and caliber and I guess since then we’ve been doubling head count and tripling revenue since we launched into the market. We were fortunate enough to bring Index and Jan Hammer on board in January. Jan, of course, has amazing pedigree having backed companies like Adyen and RobinHood, so the extent that everything that he has touched is going to turn to gold and he has an amazing understanding of teams and fintech. I think he really understands the potential for what the company can do and what we can build and I think, for me, it’s great to have that level of guidance and prescience around how to build a company, how to evolve the product, how to build a team and I think I’m fortunate to be able to work with that quality of investor and director. Peter: Sure, so we’re almost out of time, just a couple of more questions. You said it’s an arms race between the bad guys and the people like yourself. Do you think, are we ever going to get to where we’re capturing 10%, 20%, God forbid 50% of all of these people, or do you think it’s such an arms race that it’s going to be hard to really get, you know, 50 times better than what we’re doing today? Charlie: I think and I hope that what we’re building, I mean, what we’ve built today is a very small portion of what we want to build in the future and I think we can do…I think with the right technology, we can solve this problem. I think the way it’s done now is crazy, right, and I think, hopefully, with our roadmap and with our investment we should be able to make a huge dent and I think, hopefully, we should be able to stop a much greater proportion of it. I don’t think it can get much worse than it is today. Peter: So you think that long, long term. As you were just talking, it reminded me of like spam with e-mail like 20 plus years ago where it was such a problem and now, spam has pretty much…I mean, there is still some that gets through, but it’s very much a solved problem. Do you really think that we’re going to have…that AML will be a solved problem in the next several decades? Charlie: I think we should be able to eliminate it from quite a few different areas. Part of the reason fintech exists and why alternative lenders exist is because the big banks have done not a great job at their core business and a lot for their core systems just aren’t functional. So I think, hopefully, with the right investment, the right technology…I think versus…currently, it’s a drop in the ocean in terms of what’s actually stopped, I think we’ll be able to make huge strides going forward in completely eliminating money laundering and terrorist financing from huge swaths of the economy. Peter: Okay, very interesting. Last question then, what’s next for ComplyAdvantage, what’s on your roadmap say for the next 12 to 18 months? Charlie: So I think, it has been since day one just relentless, aggressive investment and improvement in the core systems. I think the incumbents haven’t managed to find a solution to the problem and they haven’t invested and they haven’t solved the problem. Given what we’re doing and given the rate of improvement, we think every day we can make huge strides and the compounding effect of those will be huge. Hopefully, we can continue our current level investment and improvement and we think we’re going to make a huge dent in a massive problem and improve both society and the economy for everyone else. Peter: Right. Well on that note, we’ll have to leave it there. I wish you the best of luck, it’s certainly a needed service. Charlie: We definitely need it. (Peter laughs) Peter: Great. Well, thanks, Charlie, I appreciate you coming on the show, Charlie: Grateful, Peter. Peter: See you. Charlie: Bye. Peter: Well that was fascinating, you know, you can hear it in Charlie’s voice how enthusiastic he is about solving this problem. As he just pointed out there towards the end, I mean, the reality is the existing financial institutions have done a poor job, they’ve done a very poor job at really mitigating money laundering despite all of the billions of dollars and tens of thousands of people that are involved in this, the results just aren’t that great. I think we need a new approach, something like what Charlie’s company is doing is so desperately needed so that we can, hopefully, down the road, maybe several decades down the road where money laundering becomes a much smaller business than it is today. Anyway on that note, I will sign off. I very much appreciate you listening and I’ll catch you next time. Bye. Today’s show was sponsored by LendIt Fintech Europe 2019, Europe’s leading event for innovation and financial services. It’s happening September 26th and 27th at the Business Design Centre in London. Registration is now open as well as speaker applications. Find out more by going to lendit.com/europe.[/expand] You can subscribe to the Lend Academy Podcast via iTunes or Stitcher. To listen to this podcast episode there is an audio player directly below or you can download the MP3 file here. (closing music)