The Power of Data Podcast
Episode 3: Using data to thrive in a complex world
Guest: Jason Crabtree, Co-Founder and CEO of QOMPLX
Interviewer: Sam Tidswell-Norrish, International CMO Dun & Bradstreet
Welcome back. You're joined by me, Sam of Dun and Bradstreet, and Jason Crabtree, the co founder and CEO of QOMPLX. Welcome, Jason.
Thanks, Sam. It’s really nice to be here today.
So we're recording this from the heart of, I guess, UK capitalism, if I can be that bold at the London Stock Exchange. We're here for the Motive Labs FinTech day, introducing large banks with some of the most exciting emerging technologies, which is why you’re here. This podcast is entitled The Power of Data. I feel like for this one, it should be the power of data and analytics, because that's what you guys do. Can you tell our listeners a little bit about QOMPLX’s mission?
Sure, I think QOMPLX was really about how do we actually help organizations thrive in the face of tremendous complexity. We're in a world that's increasingly interdependent. And there are all kinds of different sources of information that are competing for our attention. The way that I look at it, when data becomes very, very cheap then attention becomes very, very expensive, and I think our role in the world is really to help organizations capture and understand their roles in that changing environment. And to try and do it in a way where we provide a lot of the infrastructure for them. We talked about it as sort of embracing this complexity, so they don't have to and reimagining it in the context of something that's positive for their businesses. The organizations that can win the future are the ones that honestly are really good at navigating this complex, interconnected environment and figuring out how to distill insight from all of these competing sources of information.
I really like that tagline: ‘attention is expensive’. And you're absolutely right. Today, the world is speeding up, every day. John Thompson said it on the first podcast we ever recorded at Motive Partners, he said, ‘the world will never be this slow again’. And that was a terrifying thought that stuck with me an enormous amount. So let's talk for a moment about sources of information. Just explain to our listeners what the new sources of information are, because sometimes it's hard to wrap your head around how much data is being created every day?
Well, I think there's a couple different classes of information that we're trying to sort of wrestle with. One of them is that we've got organizations that are sitting on their own proprietary information. And that proprietary information might come from their different business processes from the nature of their engagement relationships with their own customers. And they've got other information that's less around the relationships and more around the operations, right? A manufacturer, or a bank has slightly different types of information than you get from a retail organization. And we go a little bit further and you start to say, well, there's all this government information, there's economic information. There's also, you know, growing types of companies out there doing all these different types of sensor capture. So whether that's private satellite constellations, whether that's folks that are listening to AIS data coming off of ships around the world, because of their safety beaconing requirements, every single one of those types of feeds that's available, you've got to do something with it. And in the context of actually surfing the tremendous amount of information, this wave that's sort of coming at you, you've got to find ways to sort of identify what is the economic basket, the collection of that data, the right mix of your data, licensed data and open data – true public data – that might be free or true public source information, and to ultimately get to some view of how that information is going to be employed into a system that you've got. But you’ve first got to be able to identify a lot of cases who did what, to whom, and when? And can you actually say that with some confidence?
Thanks, ‘who did what to whom, when’ is a really nice way of thinking about data and analytics kind of in a tagline as well. You’re simplifying all this stuff for me, I'm really grateful, thank you! So let's talk a little bit about the genesis of QOMPLX. You started in 2015, just four years ago, and not a great deal of time and today you have over 140 people, you have 60 patents out there. You have done hundreds of thousands, if not millions of hours of development work. These are big statistics for a firm that's so young. We're going to talk about talent a little bit later, but that's a key ingredient to covering so much ground and so little time. Can you tell us about the start of the journey and what sparked it?
Andrew, who is my co-founder, he and I met in graduate school in Oxford, and he had gone to the Air Force Academy and I had gone to West Point, and I ended up being fortunate enough to win a Rhodes Scholarship. And we met in grad school and I was a very lazy engineer. I was constantly trying to automate my way out of a job. And what I was typically trying to do was take this very well ordered, well-structured data and try and automate some sort of design process. And Andrew, his background was in, his DPhil [Ph.D.] at Oxford was specifically in unstructured web scale data extraction. Some quite well recognized academic work on that recognition, you know, conferences, like very large databases and others for building huge real estate data sets and other types of price check data sets and other elements. And he and I started to collaborate on before going back and serving in the military and I served as an infantry officer Afghanistan, he actually ended up doing some chief architecture work for the Air Force. And then we ended up back at Cyber Command and did a lot of outreach to systematic risk and financial services, etc. But what we kept finding across this entire journey, whether that was the academic portion of our journey, whether that was some of the research I had done even as a visiting researcher at Rice University include structure interactions or others was we wanted to be able to observe the world. We wanted to be able to capture information about who did what to whom and when. We wanted to be able to think about how do we extrapolate from the historical information, which is a lot of what people think about when they think about machine learning, and AI, even most people aren't really talking about it. They're talking about sort of retrospective modeling. And then we also want to do bottoms up simulation, we wanted to be able to look at things like agent-based modeling, we want to look at how people might interact in a future environment. And then we want to identify what information could improve a decision and go all the way back around the horn. And that ability to go from data collection, to extraction to ingestion, and schematization and normalizing data. And then thinking about doing modeling of all kinds of varieties over the top of that to be a bit more predictive, not just to get to insight, because insight isn't really all that important, actions are important and actions that are the right actions are important. So our business came out of a bunch of experiences where we felt like we maybe did automation, but we ended up with an experience or with a use case where maybe we made more decisions, we made them faster, but it wasn't necessarily doing it better. Or we got to insight but I don't want to watch a high def train wreck, I want to avoid the train wreck. And I think that's where our business really sought to differentiate itself and to start investing in linking together horizontally, more of that chain across a couple of very narrow, specific applications, to drive tremendous value to hopefully impact and have a better outcome.
Let's talk about some of that value, then because there's no better way to demystify something than to apply it to a real use case. I'm thinking on the spot now, but let's go to banking, it’s an area I know well, and I know our listeners know well. If we take a bank, a traditional bank, not necessarily a challenger bank, they're creating huge volumes of data and then leveraging huge volumes of data. Whether they buy it or they're matching it with ultimate data. And they use it across the business. They use it for loan approvals. They use it for accounting. They use it for fraud detection, KYC [Know Your Customer], AML [Anti-Money Laundering], and everything in between to get a 360 view of the customer. They're using it to grow revenues, but also to mitigate risk. Sure. How would QOMPLX work with a bank?
I think there's a couple different ways that QOMPLX would work…
Yeah that's a big question, I apologise.
Yeah, sure. Let's separate it first into kind of two buckets. The first bucket is something we're really known for, which is some of our work in cyber security. So on top of a common platform, I think this is a common theme. We do some advanced work to help them mitigate the likelihood of identity-centric breaches, how do we make sure that you are who you say you are on a network, that the logs that you've got are appropriately attributing activity on a network to that person? I think that's really important. We can talk about some security bits and how we can help you fuse together operational data, to understand the heartbeat of the institution, from the actual networks that support the business. If you zoom out and you start connecting that into, you know, something that actually drives growth, that drives value to the customer. That's all part that operational risk management sphere, and this is ‘how am I supporting the processes that I’ve got?’ If I have a new product, is it cost effective for me to integrate and service those types of customers? Am I targeting in the right way? Is it actually something that's going to add to bottom line, not just top line revenue, when I'm trying to build my growth strategy? When you think about the pure banking applications, and as you noted, things like KYC, and AML are for organizations that do have internal trading operations, and they want to be able to look at their market risk functions. And they want to be able to look at things like some of the trade surveillance issues. Well, that's where this data fusion challenge continues to persist. It's actually quite similar to the data fusion challenge they've got on the cybersecurity side, you've got a bunch of sensors across your organization and they feed data, you'd like to have a picture of who did what to whom, in your network. Well then you'd like to have that and, not at the technical level, but you'd like to have that at the business process level and effectively say, ‘how do those processes are actually linked to one another?’. And what was the way that the client interacted with me? What was the way, the path that that client took? The checks and balances in there? So we do work on both of those two categories.
We're talking about who did what to whom, and when
And we're talking about analyzing trade surveillance and client flows and all that sort of stuff. How predictive can you be? How can you get ahead of the curve and find the bad trade before it happens? The bad trader? How do you know if a client may default? How do you know if a client is going to commit fraud?
I think there's a couple of phases of maturity that organizations have to go through. And I think that's a really critical takeaway from us and our experience and variety of government applications, private sector applications and academic applications. When you want to answer those questions, and you want to do these predictive types of elements, the first thing you've got to be able to do is to even know where your data comes from. You've got to be able to understand how that data is processed and ingested. You've got to know that you've got a data model. You don't want data lakes that are not related to a data model that relates to your business use case because they're very inefficient. And the bigger you are, the more that that statement matters. So you have to think very carefully as you're constructing these technology systems or you're integrating or doing vendor selection that you are, you know, not falling into the trap of imagining that there's free lunch. And you know, in that environment, you want to think really carefully about what's the business application. And so the way that we approach this with clients is we spend a lot of time making sure we understand the question that they're really trying to answer. And some of our applications do this for them. Right? We've got some turnkey things that are specific questions like trade surveillance. But if you want to just build analytics, you got to know a bit of the question that you're trying to answer. And what we do is we try and do a good job of integrating that data set, getting it actually into a well-ordered structured environment from whatever unstructured state or semi structured state it started in. And once it's there, you know, like a graph database or a knowledge graph, or time series data split into a time series data store, you can ask a lot more questions a lot more efficiently, and at much greater scale. And so once you get that data in, then you want to start saying, Well, how do I orchestrate that data flow? We like to start simple with things that you understand. Right? I think a lot of organizations try and they want to jump to fancy words like deep learning and everything else before they can even do some basic rule-based things on their data. Rules are powerful, statistics are powerful, then you can start to embrace machine learning, then you can start to embrace things like autoencoders, where these deep learning elements are going to be a little bit more black-box, and they have trade offs. So we like to look at a blend across that. There's no silver bullet type of math, there's no God model here. Right? I think that's one of the temptations that we'd really try and dissuade, you know, our clients from thinking about is sort of pretending that there's going to be one. So we'd like to give them an infrastructure, we help them integrate that together, we help them do that in a way that's reliable, that's robust. They can write their own jobs. And we'll also sometimes help them get some of their use cases together to be able to profile behavior, profile anomalies, or do behavioral or change detection or other types of issues that are there for them.
That's a really cool point around there isn't a God model. And that's exactly at D&B, the methodology and thesis we had too. It's a fact I know, but you're creating partnerships with people that solve different parts of the problems around the value, down the value chain, is critically important and it's definitely a new approach. Not just for us, but I think for the data industry itself. Let's talk about something that makes QOMPLX unique. In another way, in a different way. It's the people right? So I, hand on heart, I didn't know where Reston was in the world. But I've been doing my research and it's, it's a pretty incredible place. It kind of feels to me like, what Silicon Valley is to venture, Reston is to cyber. And you've got some of the most world-class engineers on the planet residing with you, in Reston, churning every day. And then on the other side of the spectrum, you have some incredible business leaders as part of the QOMPLX journey as well. So talk us through a little bit as the CEO, how are you pulling together that patchwork quilt of talent, to make you guys so special?
I think that when we started the business, we really had a thesis that talent is global, and that we wanted to be able to integrate with centers of talent in different locales around the world to be able to provide the right expertise and deliver it via common platform with a very partner-oriented commercial strategy. And that was really core to the work that we do at QOMPLX. You know, we do some things under our own brand. We also do some co-branded work and some white label work with large organizations. And we found that that's a really powerful model. And part of the reason that we, I think, grew up that way, in many ways was that where we are headquartered in Reston, Virginia actually is a bit like, you know, old cities being built on rivers and ports, right. And one of the largest fiber switching stations in the world is there in Ashburn [Virginia], which is why it's one of the largest cloud and data center provider sort of areas in the United States. Just like Palo Alto and other elements, there’s sort of the topography of the internet has influenced the siting of us and other organizations. But you know, our UK team is growing in its technology and unstructured data work, but it also includes some of our insurance experts. So you know, Alastair Speare-Cole, who's an extremely accomplished global insurance, and reinsurance, executive leads our UK operation. Andrew, who's our CTO spends time circulating across our different global offices. But he also, you know, is growing a team in Denver, where he is based. Our Virginia team is where our headquarters is, so we have a lot of our operations and accounting executives and marketing and other things based there. And we've got a tremendous portion of our engineering team there and our operations center. But we've been able to tie into the ecosystem that grew up around Cyber Command and NSA [National Security Agency] and other types of elements. And I think we've seen just a robust environment happen in the Northern Virginia area which is part of why Amazon HQ2 and other things are moving into that area. It's just such a tremendous hotbed of innovation in the United States right now, for us, because perhaps of some of our global educational experiences and other pieces, we really do have a view that it's important to be where the talent is, and to be a part of the communities of interest, whether that be insurance, whether that be our work in New York and finance. We like to pull experts from the industry into our business, allow them to reimagine some of their workflow, some of their daily job, and then go back to their community of interest and use a next generation technology platform to enrich their former work and their former colleagues from our platform. And that's something, that's exactly how we approach you know, taking in actuarial and underwriting talent, that’s exactly how we've approached bringing in some trading and financial talent. And frankly, it's exactly how we've approached bringing in even the cyber security talent and thinking about how do we engage with our clients.
Young business, a lot going on, you and I have spent some great time together. You’re in London frequently, how are you splitting your time as the CEO of this business?
Well, I think we're really different today than we were four years ago. When we started, we were exclusively a research and development organization, right, we had to actually build something. And I think we had been quite quiet in our development of the company in many ways, because we wanted to make sure that we did the hard work, the engineering work, to actually be able to provide some of the solutions, delivering petabyte scale data analytics as a service infrastructure for global enterprises and globally recognized brands is not a walk in the park. It took a lot of investment, manpower, engineering to actually be able to do that. And frankly, before we talked too much in public, that was where we put 10s of millions of dollars of investment and hundreds of thousands of hours of engineering resource. Now that we have sort of moved from that phase, which was the first several years of our life, in the last 18 months, we really transitioned, and we started to win some significant client contracts with a number of recognized global corporations. And so you know, we have major global technology firms, we have, you know, Fortune 100 technology firms, we have some of the largest law firms, we have massive insurance companies and financial services clients. And we're providing software-as-a-service to those enterprises with really high SLAs [Service Level Agreements] and a white glove delivery. And that's something that most of my time today is about partnering with those key clients and the executive management teams there, partnering with some key delivery partners that are helping us scale and bring our solutions – even embedded in their own – to their client base, and then working through the broader needs of our management team to help support them and the execution of that work. We very much believe that ultimately it's our engineers, our solutions architects, our customer success managers, that are delivering the value. All my job is, is to make sure that they have what they need to do that better.
That's a great answer. Thank you. Right, we're going to talk about a little bit more career-based stuff. So who have some of the most influential people in your esteemed career been so far? Who are the people that have mentored you, been role models that have taught you some of the stuff you learned that you're putting into practice every day as a leader?
Well, I think when you look at the career that I've had, and that led me to be a part of creating an organization that I hope is much bigger than what I will do, or what Andrew and I will do. This is really about a team. And I think the time that I had in the Army, especially as a small unit infantry leader, everything that you're doing is actually about going out and building an organization that is going to work together to deliver value to meet the needs of the client, but it's about a mission. Our mission here, and I think what we found, even with some of our core investors was that we said we were going to democratize core capabilities in analytics infrastructure, that right now are really sort of restricted to some of the big wealthy corporations. And we're going to help provide that as a service to make that much more accessible, to help other organizations transform. Or even organizations that could have maybe afforded to do that on their own, do better via a partnership with us. And what we found when we started to look at partnering with Motive Partners, and certainly with Steve Daffron who has been wonderful in helping actually catalyze some of the momentum behind this business and also certainly with Bill Foley and the broader team at Cannae Holdings that was involved with us. They I think were willing to partner with us with this mission that says ‘we actually want to be a part of other people's story’. We're an infrastructure provider, we're an enabler. Analytics and data are supposed to be both ennobling and enabling. And one of the reasons why we felt like their work with us was so important, and the values that were there were so aligned with ours as a business, was the same kind of leadership philosophy with some of our core investors that we felt was about how do we embed and enrich an ecosystem of counterparties that need this type of technology to improve their own clients or their own experiences. And that I want to do that on a very collaborative long-term basis. We didn't start this company to go try and sell it. We started it because we believe there was a tremendous technology and service quality gap in the market. And that a lot of businesses were being forced to adopt a very fragmented technology ecosystem for this infrastructure that was not necessarily aligned with the values or interests of their own business. We thought something should be better aligned to financial services and other companies that needed to have this pre-integrated platform. And so for me, I've been really grateful to have the kind of trust and confidence from folks like Bill Foley, from Steve and from some of our other investors, Arnold Chase, who's on my board, Bill Murdy, who helped us get started in many ways, and say, we're going to go build a lasting thing. This is something we hope to be doing for a very long time. And if we, I believe, have that mission-oriented focus, and we continue to exclusively partner with investors that believe in the mission that we've got, then we have a much better shot at our ability to execute against that and really deliver the value we seek to provide.
I'm from an Army family and I went to an Army school, so it's a perfect analogy, right, bringing people together, setting the mission pointing at the hill and climbing it together, I think is a great way for a leader to view not just his career, but the building of a business that wants to be part of other stories. That's really powerful stuff. Jason, we've got other things to do here at the London Stock Exchange. I could talk to you all day, you know that, I normally try to, thank you for being on the podcast with us. We're very grateful both as Motive Partners for your partnership but also Dun & Bradstreet. So thank you.
Thanks so much, Sam. Really great to be here.