Turbocharged AI analytics, with Carey Anderson
Big data analytics used to require multiple sources of data, feeding your systems via a complex array made up pipes of varying structural integrity: leaks, hassles, and something to be left to only the most advanced of data organisations. In today's episode, we find out how all that has changed:
“As the name suggests, we are one big data pipe with a lot of information flowing through it, and we fuse that with AI to produce some high-value analytical products coming out the other end. And this 360 degree view of a customer, with more than 500 unique attributes, is a pretty powerful position to be in. And from there, we can do some wonderful things in real-time.”
1datapipe is at home at https://www.1datapipe.com/ and that's where you can arrange a demo
Or you can find both 1datapipe (https://www.linkedin.com/company/1datapipe/) and the wider Provenir Group (https://www.linkedin.com/company/provenir-global/) on LinkedIn
And while you're there, come and find and connect at https://www.linkedin.com/in/brendanlegrange
Meanwhile, my action-adventure novels are on Amazon, some versions even for free, and my work with ConfirmU and our gamified psychometric scores is discussed at https://confirmu.com/ and on episode 24 of this show https://www.howtolendmoneytostrangers.show/episodes/episode-24
If you have any feedback or questions, or if you would like to participate in the show, please feel free to reach out to me via the contact page on this site.
Keep well, Brendan
The full written transcript, with timestamps, is below:
Carey Anderson 0:00
I mean, at the end of the day, we're all about the customer journey and the 360 degree view of that customer.
The deep analytics and insights that you can yield from data when you apply certain methodologies, technologies and models - and the combination of alternative data and AI working together in unison with datasets that the bureau has - to deliver something powerful: that's kind of how 1datapipe was born, with a vision of serving emerging markets to solve the problem of financial inclusion. You know, 70% of Latin America could be considered underbanked.
Brendan Le Grange 0:39
At the end of every episode, I encourage you to reach out and connect with me on LinkedIn. And from time to time listeners do, which always makes me happy. And on very rare occasions, I'll even hear from a listener in a work meeting, which makes me even happier.
But once, and only once, I've been recognised by voice. Now, I assume Mike Holmes had known I was at the University of Edinburgh's credit scoring and credit control conference, I'd publicised my attendance on this show, and perhaps he had just been looking for me, but I'm willing to round that up to being my first real celebrity experience.
But our subsequent conversation has created something of a dilemma for me.
Pulling in one direction is the fact that I learned he listens to the show while swimming at the gym. So if an episode runs over about 20 minutes, every extra minute is risking him either overtraining or, worse, drowning!
Pulling in the other direction is the fact that this episode features his boss, who is talking about cutting edge AI applications, about improving access to credit (especially in emerging markets), and about data-driven lending strategies. Three topics I just can't hear enough about.
So stay tuned right to the end. To hear how far we push Mike today. Welcome to How to Lend Money to Strangers with Brendan le Grange.
Carey Anderson, CEO and co-founder at 1datapipe, welcome to the show.
Carey Anderson 2:20
Thanks, Brendan. It's good to be here.
Brendan Le Grange 2:22
Carey, when I started my career, around the turn of the century, I was just starting to become aware of terms like AML and KYC. But for me, I think of KYC being a 2010s plus phenomenon.
But actually looking at your LinkedIn, you were well ahead of that curve: already Director of Global AML and KYC back in 1999. So before we talk about what you're doing now with 1datapipe and AI, would you mind starting a little bit further back in your career and just talking to me about how you got there?
Carey Anderson 2:56
Yeah, sure, Brendan. I grew up in Melbourne and then, after I finished my university degree in economics, like a lot of Australians, I went to London, and started my career there - accidentally in data at a company called 192.com. We were involved in all different types of data, but really what kicked it off for me, that led me down the identity verification path, and KYC was, I guess, the events of 9/11.
After 9/11, the Patriot Act was introduced and from that, we started hearing terms like anti-money laundering and counter terrorist financing. And soon after that, you know, know your customer became a pretty common term. And one of the requirements was electronic identity verification.
So the data that we were using, combined with the electronic verification requirements under KYC became the base of of a solution that we developed in 192.com was one of the pioneers in that space. So that led me on my path to electronic verification in general, which then opened the doors to many things later in life.
Brendan Le Grange 4:06
And some of those many things were an entry into entrepreneurship - several times over, actually. And so I'm not going to try and parse out all your many ventures, but maybe we could just talk a little bit about that: what inspired you to move from the life of an employee to your first startup?
Carey Anderson 4:24
After 192.com, I jumped across the pond to Washington, DC where I worked at integrity. And you know, I saw some really good things happening. And I saw markets opening up and I saw this whole industry becoming a very global phenomenon in terms of people having to have these infrastructures in place in every business to be able to understand their customer and safely onboard them.
So my experience told me that from the UK to the US, I could maybe do my own version of this. So I went back to Australia and I started up Global Data Company. We were a very small company to start but with a very big global vision. So we were the first company, I believe, to reach 50 countries that could safely onboard clients under the KYC legislation. And that sort of put me, I guess, on the global stage when it came to identity verification.
And you know, really were focused on that for 10, to maybe even 12 years. And at a certain point in time, being young, we got to tap on the shoulder and someone offered us at the time what we thought was a lot of money. And so you know, we did an exit, and probably too early, to be honest, because that product, which was called Global Gateway, went on to be purchased by Trulioo, who were probably the biggest company in the world today for client onboarding and KYC. That product, Global Gateway is still their foundational product.
But zero regrets.
And you know, that led the way to other things moving forward. And then I went to help LexisNexis Risk Solutions, and had a really good time doing that and enjoying the consulting aspect. And then I got to the point where I wanted to get back into the game.
And that sort of leads me down the path to my introduction to Lauren Smith, and how we collectively decided that we were going to do a project together. And that's really the beginning of 1datapipe.
Brendan Le Grange 6:15
Yeah, and 1datapipe, where insights become decisions in seconds. So and we're gonna get into the nuts and bolts about what that means and how we you apply cutting edge AI to make that sort of speed of decision possible, but before we get into that, let's stay at that high level for a moment. What is 1datapipe? And where do you fit into this quickly, evolving landscape?
Carey Anderson 6:37
So as the name suggests, we are one big data pipe with a lot of information flowing through it, and we fuse that with AI to produce some high value analytic products coming out the other end.
You know, we're looking at big data and lots of information happening in real time and delivering scores that have some impact on being able to make a decision in rapid speed. So I didn't even verification is a component of it. But it delves much more into the deep analytics and insights that you can yield from data when you apply certain methodologies, technologies and models.
And so what I wanted to do in this venture was look at the bigger picture and not be so pigeon holed with products that were compliance related. And the other part of that was that I'd been intrigued by emerging markets. And most of my experience prior to this was in established economies, the UK, Australia and the US, I wanted a solution that would fit into Latin America, Southeast Asia, and even MENA eventually.
And that's kind of how 1datapipe was born to solve the problem of financial inclusion.
Brendan Le Grange 7:48
In the last few decades, if you were at the cutting edge of regulation of IT capabilities of analytical tools and techniques, you were serving the biggest lenders in the most established, most developed markets, and even then only when they dealing with their more sophisticated clients.
You were limited mainly to the tip of the pyramid in the first world. But now we can apply some things that are really cutting edge from a technological point of view, into developing markets around the world. And as you said, for 1datapipe. That's not just coincidence, it's a core part of the philosophy.
So maybe I'm going to jump out of the narrative here and talk about that. If you think about 1datapipe's global reach, and this bigger issue of how AI is helping increase financial inclusion - how is that different compared to what it might have been just 10 years ago, were you're trying to solve the same sort of problems? ,
Carey Anderson 8:45
Yeah, I mean, back in the day, you know, we were doing a lot of data matching and a lot of manual intervention in some of that modelling and then meshing together huge datasets to form these solutions for identity verification purposes. And the complexity of dealing with such large datasets with limited tools was one of the challenges at the time. So I guess moving forward, a great number of those problems anyway, have disappeared now with the introduction of AI and the introduction of other technologies that support being able to synchronise and being able to model this unstructured data and do things in lightning speed that would have taken us weeks, if not months, back in the day.
So there's been a dramatic shift.
Brendan Le Grange 9:29
Yeah. And I think back to one of my earlier jobs, maybe it was 15 years ago, there was a project where we needed to do some analytics because some some numbers had gone awry. And one of the genuine suggestions of how to get the data we needed was that there was a room upstairs that was filled with printouts of the old credit card statements, and to be going through those and kind of capturing data from slips of paper, and then trying to build databases.
And, yeah, 15 years later we've leapfrogged all of that.
And we don't have to worry about that being the prospect we face if we go into a new market, which, yeah, I think it's really exciting. And just as a global industry gives us so many more areas where innovation might pop up, that's useful for us. Because equally it means that things that work in emerging markets can benefit develop markets to.
But let's go back to 1datapipe, as you said, you built that with Lawrence Smith and this is powered by Provenir AI. So people who may not yet know the 1datapipe name, probably know Provenir, how did that come to be? And and how do these two businesses interact?
Carey Anderson 10:39
Yeah, so 1datapipe is part of the Provenir group.
And when I met Lawrence, or Larry, as everyone refers to him, you know, he had a strong passion for AI. And I had a strong passion for data. And so we shared a similar vision in what we wanted to do, which was a big project that reached global markets amd that had a combination of alternative data and AI working together in unison to deliver something powerful.
At that time, we weren't quite sure of exactly what the outcome would be, but we felt comfortable that we're on the same page enough to kick off in the start at '23.
And the relationship really has been wonderful for me as the CEO, because Lawrence opened up the door to Provenir, in terms of all their wonderful resources from legal and from financial accounting, and obviously, the AI team that he'd already had existing in property - so really a plethora of wonderful resources.
So when I looked at the project, it wasn't really a startup, it was really an extension of the property business to a degree with my flavour, and Lawrence's flavour. Our actual real name is Provenir Data Inc. but we trade as 1datapipe. And you know, we're the sister company within the Provenir Group. And, you know, a lot of thanks goes to Larry for for supporting that initiative.
Brendan Le Grange 11:57
1datapipe is highly complex AI, customer analytics simplified. So a hugely big ask - what does that 1datapipe product look like?
Carey Anderson 12:08
I mean, at the end of the day, we're all about the customer journey - and this 360 degree view of that customer with more than 500 unique attributes that belong to an individual is a pretty powerful position to be in. And from there, we can do some wonderful things to create products, you know, we look at our solution in terms of layers, really, before we even get to scores.
And at the start, we take the identity of someone, and we look at if that is secure, or if he is prone to any fraudulent activity. So the first layer is secure ID and fraud. And the data points can include things like digital footprints, web screening, phone, email, IP risks, other stability, behaviour controls that we stare at, and we analyse. We try to prevent that fraudulent activity that happens at the start at the customer journey, things like account takeovers, or stolen identities, synthetic IDs, which are very prone in the market today, individuals that have no intent to pay for the ID and the fraud screening where we move to someone's income, we'll look at how stable that income is.
And it's very difficult in some of these emerging markets to obtain access to some of that data. So we have a combination of actual data that we source from government and different data providers. And then we infer some of that other information. And we use things like actual income, household income, someone's job history or their tenure, we also take into account things like people being paid in the gig economy.
And we also model all this data points against the national average of what that job title looks like in different geographic areas so that we can get a median range, and factoring things like social security and insurance into those. So I think we've done a really good job in the solution. In fact, an excellent job of income estimation, in my opinion,
Brendan Le Grange 14:08
I just actually saw this week on LinkedIn, and I forgotten who I saw it from so forgive me for not attributing it, but somebody was talking about how, in India, it's still standard practice to say, if you haven't worked at the same employer for three years, then you won't get credit because you're showing volatility and, those sort of across the board rules. We're leaving so many people out. And even when you included them, that's three years of one payment slip.
But you've got all the rest of it to try and consider particularly when you're talking multi generational households. It was a wildly complicated problem that we would often just say, No, I don't know how to solve the problem. So we're just going to put in place some some sort of really rigid rules.
Carey Anderson 14:50
Yes, certainly. And like I said, at the backbone of any decision is fundamentals and that income plays a very important role. For example, in Brazil at the moment We have data on 99% of the adult population, we've done some really clever modelling. And when it comes to income estimation, we are able to make very accurate predictive models on income, which hasn't been done before. Where it's highly accurate. And, you know, we feel like that's a massive leap forward in financial services.
Brendan Le Grange 15:19
Yeah, I mean, you'll know this as well, but I don't think that's been done in the US at that sort of scale, either. So this is one of those examples, I think of technologies and use cases that can be applied, going back into the big developed markets as well.
Carey Anderson 15:32
I agree. And then, you know, from the income, you sort of delve into the credit assessment layer. This lazed is a little bit different from region to region, you know, so in Brazil, we actually work with the credit bureaus, there is quite a big participation rate compared to other regions of the world, believe it or not, especially after COVID. So when it comes to credit assessment, we use a combination of alternative and traditional credit modelling.
So we actually partner with bureaus in country in Brazil, to be able to access some of that traditional credit data.
But in other parts of the world, like Indonesia, we were partnered with the MNOs the telco providers, and we'll be using those telco data points to be able to calculate someone's credit assessment and credit risk. There's a wealth of information outside of the bureau with these days that help find those people that are being filed credit if you live off the grid to the credit bureau, which is still a great majority in many parts of Latin America, and it's still a very big percentage of of what we're calling underbanked, or, or underserved in some countries.
But within that credit layer, we're looking at an individual's mortgage, or their rent, their mobile phone payments, their utility bills, do they have credit cards, do they have retail store cards, have they been in the collections process, combine it with datasets, the Bureau has to form that credit assessment.
And that's how we form the next part of the solution, which is really the Credit Trust Score, being able to tell a bank or a FinTech or any financial institution that you've missed out on this opportunity, and this opportunity is this individual that we can tell you more about is something that hasn't been done before. So we call it the Financial Inclusion Score.
But you know, really, what we're doing there is we're looking at all of the profiles across Brazil. So taking into account the entire population. And we're clustering that, that population into six to seven different groups. And we're looking at the behaviour of all those different groups through our modelling and through our analytics, looking at things like their lifestyle, and we're comparing them to lifestyles that are similar in that group, and looking at other groups looking at the distance and the gap.
And that's a way in which we can determine that that individual is capable. And once they've gone through our customer journey, like I mentioned, at the start layer one, layer two, layer three, once it met all those criteria, and then we can finally determine that there is a gap, meaning there's a propensity to consume more, we can then calculate that person's financial inclusion score and tell the bank that this person isn't fraudulent is safe, does have a good income, does also have a credit assessment, and is in a position to consume more financial products.
And the banks have not had that information to date. And that's what we call the financial inclusion score, which we think might be a game changer, when it comes to connecting the underbanked to the financial products and services that they need to thrive, you know, they can better their lives, new markets open, more competition is available new products and exciting vendors that feed into these products spawn, and you know, the industry should come alive, especially for emerging markets that have been laying dormant for too long, because there's just been no data and no analytics on really individuals that are somewhat qualified. And if not somewhat qualified, and maybe very qualified. They're just not on the radar of these financial institutions.
Brendan Le Grange 19:10
Because the old model would have been, well, I know nothing about this person.
So my options are one to build a product that's really limited in exposure, and then we can all look and see how they perform or two is just saying Well, fine, let's take a gamble with your saying here's a fully described person, build them a product building many products serve their needs, you can provide all the services to the consumer from day one, and that acceleration of competition and all the benefits that brings.
Carey Anderson 19:47
Yeah, I agree, I think is a game changer.
And I think what's interesting as we as we looked at the financial inclusion score more we realised how important lifestyle was as well as behaviour. And that sort of led us down to something were developing to the moment which is really based on geographical havior and customer blueprints for more targeted marketing strategies, we derive a lot of this information directly from the mobile, someone's behaviour on that phone and their choices and their lifestyle patterns and gleaning all that information from the mobile, which is all anonymized data at one point.
And what we do is we take that anonymized data and because of population data that we have, and the national coverage that we have, we can tie them together and geographically link that information. So we really understand the populations lifestyle attributes better. So now, you know, once they're in the organisation, how do we do cross channel mapping for the customer journey? How do we provide a much more tailored experience for that individual?
And I think that's only going to become more interesting as we get access to more of this. We call it infinity data. ,
Brendan Le Grange 20:51
Yeah, well, I think it's also worth underlining that factors who the name suggests there's two, but one data pipe, you also understand and bring the data. So this is all about taking the data and also then turning it into decisions.
Carey Anderson 21:04
Yeah, I mean, we certainly need to have products as a company, because that's how we achieve revenue. But we certainly also want to have an open mind and be very analytic focused, and very customer focused. So where we are today, you probably won't be where we are in 24 months, you know, we'll those products will broaden. And those products will also be more tailored towards the customer feedback that we get in both those regional markets, Latin America and Southeast Asia. B
ut at the end of the day, a score is a way of wrapping all that information around to simply deliver it to an institution that can then act on it. But behind that score sets an amazing amount of analytics and AI that is doing wonderful things that we haven't been able to do up until the last few years.
Brendan Le Grange 21:48
I want to pick up on that, or just sort of benefit from your experience of building this AI-based company, you get to see all the various options and all the ways and try out different ways of applying AI to new problems. So from your point of view, what is AI is role in moving us beyond the sort of traditional data centric paradigm that we've been working in?
Carey Anderson 22:10
Good question. And I know we're all grappling with the concept of AI and how it impacts us in different ways. And business is no different.
It plays an integral role with us, at the end of the day, what what it's done for us is accelerate real time insights. And it certainly allowed us to advance the capabilities of general analysis, interpretation and rapid decision making, pattern recognition and analysis, being able to, I guess, have a deeper understanding of the information so that we can make an informed decision.
And leading into the predictive analytics, that's really where the machine learning kicks in. Being able to understand trends and behaviours from not only current, but historical data is really a leap forward, especially when it comes to lots of unstructured data, like text documents, or social media posts, customer feedback forms, just really looking at a holistic view of infor information across the board, what we call Advanced Data Processing.
And you know, that sort of delves into deep learning, and being able to process large volumes of diverse and unstructured data to really personalise that customer experience, through the products.
One of the areas that I think AI has done a really good job where we didn't do a good job as human intervention was detect anomalies, you know, some of those anomalies had just literally not been on our radar and the irregularities that are across the board and data weren't visible to us previously. So having less reliance on that manual intervention, and having more reliance on things that can pick up those anomalies, has put us in a whole different ballgame. And you know, many people will also say that it's just enhanced data security in general, across the board.
So I guess, at the end of the day, you know, we've got better analytic capabilities, we've got better predictive modelling, we've got more automation, and we've got the ability to extract richer insights.
Brendan Le Grange 24:07
Carey, on www.1datapipe.com/contact you talk about transforming your metrics in under two seconds. That's a fundamental shift, and it may even sound slightly impossible to some people listening. But you back that up with experts I see available to chat and answer questions, but also the ability to book a demo.
So if somebody is listening, who would like to learn more about what you're doing, but in particular, we'd like to see with their own eyes how this happens, and what might be possible that they couldn't imagine just a while ago. How do they go about getting a conversation going about starting this process to see it for themselves and learn more?
Carey Anderson 24:48
Yeah. And you're right, the best way is is to see it and witness it yourself. When we talk about this it It all sounds very powerful and very futuristic, but this is actually a live solution. everything we've been talking about is is in practice today as all through a single API, which is even more amazing. And when we talk about a two second decision, that's an SLA that we commit to in contract.
So everything we've discussed today from layer 1-2-3-4-5, and even the integration into any bank, big or small, is less than 10 minutes. As long as someone can digest the API and the input, you know, the solution can be running in less time than it takes to watch the regular news.
It's incredible the amount of speed to talk to one of our consultants, we can set up a demo. And more importantly, we can do two things, we can either do a batch test so that clients can have a look at the coverage, the depth, the attributes, and the scores themselves. And then more importantly, we can set up an API test where we can have a live link going within, like I said, 10 minutes to your organisation where you can query the data, you can pull back the profiles, and you can also receive the scores. And we can then discuss how that impacts different use cases within your organisation. And you know, what you're trying to achieve out of the solution.
But at 1datapipe, you know, I think the most important thing is to get the demo underway so that you can witness everything that we've talked about today and see it and feel it for yourself.
So we're in Brazil at the moment. And we launched in Indonesia on February the 14th. And then at the end of q1, we launch in Thailand. And at the start of Q2, we launch in the Philippines, in Q2, we also launch in Mexico, and Colombia, and in Q3, Vietnam, and Malaysia. So we have a lot of focus on Latin America, and then even more focused on the Southeast Asia region.
But our roadmap does extend to other countries beyond those two regions. And you know, we're excited to see where we end up in the next 12 to 24 months.
Brendan Le Grange 26:56
That's awesome. I have very fond memories working in the Philippines, in Thailand and Malaysia. So I'll reach out to some of my friends here to make sure they're having a look.
Carey, one last thing before I let you go. We are recording this in January. So it would be remiss of me if I wasn't to ask you about plans for things to come: 1datapipe is a new company growing quickly changing quickly, what's on the horizon for you beyond the market expansion that you already mentioned,
Carey Anderson 27:24
The one thing that we want to foster is a very strong team collaboration, we have a diverse range of really highly intelligent people in the company. And we want to keep growing that fundamental basis of only recruiting the best.
And I said earlier that we wanted to have an impact in emerging markets, I'd really like the company to involve itself in some of those regions outside of the revenue and the product discussion points, and be involved in some of those more community based efforts. Because it's all very well to say you're going to do one thing, but you know, most companies are there for one reason, and that is to earn revenue, I don't want us to be known as just that focus, I'd like us to have an impact and you know, give give a bid back in both those regions of the world.
Brendan Le Grange 28:09
Well, I'm looking forward to seeing how you do that on both fronts. So many lenders for so long, have been wanting to expand access to credit, they just didn't have a reasonable means of doing it at scale. So I think this really could accelerate change make a big difference for lenders for borrowers and the communities they obviously operate in.
So yeah, Carey, thank you again for making the time and I wish you the best of luck in the new year as you take on all the work to expand around the world.
Carey Anderson 28:36
I really appreciate you taking the time to reach out Brent and
Brendan Le Grange 28:39
And thank you all for listening.
Please do look for and follow the show on your favourite podcast platform and share the updates widely on LinkedIn where lending nerds are found in our largest concentration. Plus, send me a connection request while you're there.
This show is written and recorded by myself Brendan le Grange in Brighton, England and edited by Fina Charleson of FC Productions.
Show music is by Iam_wake, and you can find show notes and written transcripts at www.HowtoLendMoneytoStrangers.show and I'll see you again next Thursday.