Solving Africa's Credit Gap with Open Banking and AI

Thanks for having me, Ian. I'm doing good. How are you today?

Good question. So E-Doc is a fintech AI company. We basically have an open banking platform where we integrate directly with commercial banks, especially in Nigeria, and we can pull banking data directly from these banks, whether it's for lending decisions or for travel visa applications, which you probably don't understand in the UK because you don't apply for visas. But over here, a big component of applying for a travel visa is submitting a PDF bank statement. There are a couple of challenges with those manual bank statements. The first is that they can easily be manipulated. The second is the time and effort it takes to understand them, and it's very subjective, if I gave you my bank statement and both of us analysed it, we'd come up with two different outputs. You're looking for keywords like salary, bonuses, allowances. Sometimes it could show "SA" or "SAL". And then when you start talking about SMEs, which make up a huge portion of our population, they have inconsistent income patterns and don't have these keywords, so it puts them at a big disadvantage. Then lastly, most people in our part of the world have multiple bank statements, meaning they bank with multiple banks, and all of those banks have different formats. So it makes it very difficult to get access to credit. And with the way the world is evolving now, we're seeing a lot of demand for financial data for RegTech, whether it's for virtual assets, for AML, or for CFT. So that's essentially what we do, in a nutshell.

100%. It's very difficult to get access to credit, because we don't have a credit system in the same way you have in the UK. In the UK you have credit reports, and even where we do have them here, your credit report tells you a willingness to pay. In our part of the world, we need to know your capacity to pay. And the only way to know your capacity to pay is your financial data. The reason we need to know it is because that's what influences decisions in real time here. I want to know if someone doesn't have an income anymore. I want to know if they're a new risk. I want to know if somebody has started gambling. I want to know how their spending pattern has changed. And the only way that happens is by looking at their financial data. Because we don't have consequences for credit reports, in the UK, if you don't pay your credit, there's a consequence, and that can affect your credit. In our part of the world, I need to know when someone's going to default, and I need to know before it defaults, because once it defaults, we don't have access to recouping those funds. So that's one of the issues we have here.

100%. It's giving them access to credit. I'd say it's alternative data, it's not necessarily alternative data, but it's a different way of looking at data, because what you're getting is tamper-proof financial data. So it's not only that we're giving you tamper-proof financial data directly from source, which hasn't been altered; it's also that we run it through an AI model that lets you know their income, their spending pattern, their expenses. So we give you access to a user's financial data in three seconds. Compare that to someone having to go to the bank, get a manual bank statement, upload it, and then the time and effort it takes to understand it, that could take anywhere between four hours and 48 hours. But we're doing this in three seconds. And that allows you to scale, and to make it an automated decision rather than a judgmental one.

Yeah, 100%. I used to work for HSBC, and my co-founder and I were among the early people working on open banking there. And funnily enough, it wasn't how we saw it that it's evolved into. We used to think about how we could use open banking to solve access to credit. Actually, not even access to credit, at that point, we looked at travel visas, and from a different perspective. It was more around family members asking, "Can you help me? I want to get a visa, can you help me with a bank statement?" And you give a family member your bank statement, first of all, it's very personal; they see the funds in your account, they leave the paper lying around. So we were thinking about how we could take that bank statement and submit it directly to the embassy, without the intrusion, with only the embassy or decision-maker having access to it. So it started around travel visas. But as we started working on it, we realised we were sitting on a gold mine, because there were just so many use cases for financial data. People say data is the new oil, it's not just a phrase, you see it right now. Netflix: data. Uber: data. Everyone's sitting on a huge gold mine of data. That's why Netflix can recommend movies to you, and recommend the movies people in your area are watching, because they're sitting on a vast amount of data. So we're able to take banking data and turn it into something tangible. And more importantly, if I get your bank statement, I can paint a financial picture of you: what you've been doing, where you eat, where you shop, where you buy your petrol. And what you find is that most people have patterns they don't even realise. For instance, you probably buy your petrol from a particular petrol station, not because you intended to, but because it's close to your house or your office. We're able to see those patterns, and when people go outside the norm, we're able to flag it.

I think it's a conscious decision, and if anybody tells you it's a conscious decision, I'd argue with them. For us, it wasn't conscious. Things just evolved, and they're all painful lessons you learn. The first sign we knew we had to change was in the early days. I think we were only about six months into E-Doc when we got an opportunity to go to New York and pitch. It was at one of the biggest law firms in New York, they had a room full of investors and we pitched this idea about travel visas and how we were going to use banking data. We finished the pitch, and everybody in the room thought we were crazy. And we realised quickly why: they don't apply for travel visas, that's the first thing, and secondly, when they do, no one asks them for their bank statements. I remember someone said, "Thank you so much for educating us." And we realised we didn't come here to educate you, we came to source funds. So that was the first time we knew we had to pivot, because we realised travel visas were more of an African problem. Then we started asking what other use cases we could look at, SMEs. And now we're looking at financial data for RegTech, mostly virtual assets. And it's not because we planned this; it's because that's where the regulation is evolving. There's a growing demand. Someone says, "Can you do this?" and you think about it. Look at Nigeria right now, we didn't plan for virtual assets. Somehow the central bank realised the virtual asset market is a $96 billion market that they have to regulate. To regulate it, you need financial data. And guess what, we have access to that financial data. So now we're looking at integrating virtual assets onto our platform. So for founders, I'd say you have to be flexible, you have to listen to the market, and you're probably going to pivot 20 or 30 times. You might pivot twice a day, because there's so much happening. You just have to choose where you think the opportunity is, where you think your strength is, and follow the market.

Good question. I wasn't aware we did it either, I'm joking. We just started looking into it. As I mentioned, we realised the central bank is now looking at regulating virtual assets. And one of the things we have is access to banking data from commercial banks. So that allows us, from an origination point, to understand the flow of transactions into virtual assets. That's something we can look at, whether it's for compliance, for AML, or for CFT. It's early days for us, it's something we're piloting currently. So it's still in line with what we do. We haven't pivoted to stablecoins, we're not doing crypto; we're still using financial data, just for virtual assets rather than for lending use cases.

I pretty much worked for every single UK bank, across products that have impacted every UK household. I come from a consulting background, I've worked for KPMG in Nigeria and the UK. When you do all those projects, Bank of America Merrill Lynch, electronic trading; HSBC, open banking and lending; FX platforms at Barclays; FX at Metro, you don't realise how much of it is going to come back and impact you later. But you wake up one day and realise you have vast experience and a lot of knowledge, and you think about how you can repurpose it, especially for our region. So, some background: when we started this open banking business, we actually have an open banking licence in the UK, it was more UK-focused. And then I happened to come to Nigeria and spoke to a bank. They liked what we were doing, and before we knew it we had three banks on the platform, and it grew to 13 in less than a year. We thought, there's more opportunity here, because people are more open here. So that's how we shifted. Again, nothing we've done has been because we're the smartest people in the world, it's just because we've been able to identify where the opportunities are and been lucky enough to take them. That's what we're doing with virtual assets. And all the experience I've gained across the banks now helps, because I understand how to fix problems, how to understand opportunities, how to know when to be flexible. And I think being able to work in a structured manner is even more important, having a target, a deliverable, and being able to execute. I'd say the best thing about our skill set is being able to execute. So we know we want to go into virtual assets, we have a plan, we have a roadmap, and we're going to follow it, not blindly, but to a certain level of belief.

No, consultant with KPMG, management consultant.

I think it's two ways. One, I've always had an entrepreneurial spirit. And secondly, I have a quest for knowledge. I got to the point of working for the banks where I thought, okay, I'm coasting now, I kind of know this, I'm only operating at 20 or 30% capacity. So how do I get myself to work at 80 to 90% capacity? I had all this experience, and I like things that are challenging: how can I solve this problem? People say, "I love it when a plan comes together." Yeah, you love it when it comes together, because it came together. But what happens when the plan doesn't come together? A plan comes together because you're just lucky and it works. People say you only need to get things right once, and that's the truth, you could try a hundred times, and it works once, and everybody sees you as a genius. But they forget the 99 times it failed. So the point I'm making is that it's pretty much the experience, and it's somewhat intentional and somewhat opportunity as well.

Yeah, you are saying that correctly.

100%. That was a learning curve for me. People don't understand that B2C and B2B are two separate value propositions. When I see people do both B2C and B2B, I don't know how they do it, I think it's too challenging, because you're looking at two different customers. When we did the consumer-facing product, we realised how difficult it is, because B2C is quite challenging. You need a strategy; it's not just having a good product, it's getting that product on the shelf, in front of customers, and convincing them to use it. So B2C is more challenging, in my opinion. So we decided to focus more on businesses, but that doesn't mean B2B doesn't have its challenges. It does, because now you're looking at a longer sales cycle, a cycle that could be 12 to 18 months. Then you're asking, how do I cut that 12 to 18 months down to four months? And often you don't have that luxury, because there are internal dependencies you don't control, but you're always looking at it. Initially we looked at the private sector; now we've moved to the public sector, and you find the public sector has a longer sales cycle. So now you're asking, how do I build a hybrid? To survive, I'll go for the private sector, while where I need to scale, because the large lenders today are government intervention funds and other government agencies. Those are the guys doing a million loans, doing huge numbers, but they have a longer sales cycle. So you have to work with the private sector to make sure you have income coming in while you focus. That's how we've kind of survived.

I think it was the right decision to be early. At the time, I didn't think it was, in hindsight it looks good, but at that point we were struggling. One of the two things we were struggling with was educating people. It's difficult building a product; and secondly, you're trying to educate people and sell the same product, super difficult. But we've been lucky now that there's huge adoption of open banking across the globe, whether it's India, Brazil, even the Middle East, and Nigeria is now at the forefront in Africa of launching open banking. The fact that we're in the conversation, and known not just for open banking but for implementing open banking, I think that's where we got lucky. I always say you need strategy and luck, and we got lucky because open banking just happened to be on everybody's mind now, just like virtual assets are on everybody's mind now. Luck is, as they say, preparation meeting opportunity. So we were prepared, and the opportunity came.

That's the million-dollar question. I remember when we used to go to events and everybody would ask, "What's your biggest challenge?" The first one we'd say was regulation, because in our part of the world you don't know where the regulation is going to go. However, you can also predict where it's going to go, because everything we've ever done has been adopted from the UK. Look at our NDPR, the data protection regulation, it's pretty much adopted from the UK. So we knew there was going to be a lot of UK influence. Also, across open banking, the UK is one of, if not the most, successful open banking frameworks in the world. People forget the UK actually borrowed its open banking framework from PSD2 in Europe, but they've been more successful at it. So we knew there was no way people weren't going to look to the UK for influence. That's why our open banking platform, to this day, is built along that UK framework, which is one of the most secure. When everybody wasn't looking at consent, we were building consent; we were putting all these rigorous controls in place. And that's part of the education piece, a lot of people don't want controls because they don't understand them. It's a guessing game, because you're guessing the market's going to move in a certain direction, but you also have to be flexible if you see it's not. I'd say we had about 60% certainty they were going to adopt the UK framework.

I think any product you want to sell today is a working assumption, right? Take your podcast. When you started, it was a working assumption that this was going to work. Even though you have vast experience from Money20/20 Europe, huge experience, and then went on your own, you were guessing you'd get the audience, guessing you'd get the right people to come on board, and guessing you'd get the right commercial model. So it's a working assumption: you try one way, and if it doesn't work, you're flexible and you move it. You can never be 100% certain on any product. There's this thing I look at, be careful, because Coca-Cola only sold 12 bottles in its first year. And even if you look at the history of Coke, it used to be a medicine; it wasn't sold as a soft drink, it was for medicinal purposes. Look at where it is today. Things evolve. It's a working assumption, and that's all you can work on. That's why you have a pilot, why you learn from your pilot, why you refine your product constantly. I tell people every day, Apple does releases every single week. If Apple can be flexible, with millions or billions of iPhones and iPads and Macs, releasing every week, who am I to say I'm not going to be flexible?

Yes and no. I wouldn't go as far as to say I'm influencing it, but we do train the Central Bank on open banking, in terms of sharing our knowledge of what we know. They've issued the framework and the guideline, but we've actually implemented it, we implemented it in the UK when no one knew what open banking was. Even in the UK, when it was implemented, when we were working for HSBC, it was trial and error, because it hadn't been done before. So it was a guessing game there too; nobody knew the formula for implementing open banking. So we're able to share our knowledge, we're in the room, we're training the Central Bank, which is getting ready to implement open banking right now. So in that aspect, we're able to share our knowledge. I wouldn't go as far as to say we're shaping the discussion, but we are in the room, we are in the conversations, and somewhat influencing how it should be done in the right and correct manner, especially trying to localise it for our part of the world.

Not for E-Doc, it wasn't for us. But, essentially, they stopped issuing licences for fintechs around payments, for people who wanted to do payments. We tried to apply for a licence and it was suspended. There are pros and cons for everything. Without criticising, because I do work with the Central Bank, so I have to be careful not to criticise my client, they needed to do some cleaning up. I think they'd issued too many licences, and a lot of them weren't in use, so it was a valid reason to clean up. But it also means new entrants who actually need the licence, who are actually doing the work, can't come into the market, you're punishing the innocent with the guilty. You've issued too many licences, you need to clean up, and you're now forcing people to partner with existing players. And that has a huge cost, because the price of licences has gone up, and the price of partnership has gone up, which means a huge operational cost. Those are the challenges.

So we work with one of the largest banks that do a lot of intervention loans. One of the challenges they have is identifying what's in a bank statement. As I mentioned, you're looking for keywords like salary and allowances, but that's only for salaried workers. For SMEs with inconsistent income patterns, they're not doing that. How do I know? I could do a podcast today and not do another for two months. You could do two this month and five next month. How do I know there's an income? And in our part of the world we do a lot of transfers, I transfer money to people, I can go to the coffee shop and the POS machine isn't working, so I have to transfer. And a lot of SMEs here use the same account for personal and business. So how do you differentiate what's personal and what's business? What we use the LLM to do is look at patterns, we look at clusters, we look at embeddings. We use large language models not as a text base; we used text to baseline it, we used k-means, and now we're looking at patterns we can recognise. Because we have access to huge amounts of data, we can look at whether there's a certain time, a sender name, a recipient name, a pattern. Maybe Ian gets paid about £500 for every podcast. The first one said "fee for podcast," but subsequently those texts aren't there, we're able to recognise those patterns. We can look at whether somebody gets a certain amount at a certain time of day. So we look at clusters and embeddings, and we're able to flag risk and use it for risk scoring. I mentioned a lot of governmental loans, I won't mention the agency, but there was an agency we were working with; they put out an application and got a million applications, and could only approve 200. Because even with a 12-person team, how many of those bank statements can you look at? Some of the challenges, even before you get to the LLMs: people send password-protected PDF bank statements, or the wrong dates, and that means someone has to look at each one. But an automated system cuts that out. You can get that one million down to 500,000 just by looking at the thresholds and the dates, before you even eliminate anyone. So more people now have access to get risk-scored and get that loan, versus just approving 200 out of a million.

You're dealing with banking data, and people are sensitive about it, the banks are sensitive about it. But we use consent. So first there's the education piece: people understanding that we can't pull data without consent, just like in the UK. Then there's data encryption, whether it's in flight or at rest. Then, with AI, there are biases, there always are, because you build your model a certain way. But we're constantly retraining our model, and we use different models; we don't use the same model for personal and business. Sometimes we have cases where people want to run a gender-based intervention fund, so you have to tweak your model to fit that. When I say bias, if I'm in Lagos and the bank statements I'm receiving are all from Lagos, and then I apply that same thing to Kwara or Oyo, which has a different demographic, a different gender base, a different income and affordability, that could lead to bias. So you often have to blend it, retrain your model. That's why we come up with clusters and embeddings, because you're constantly retraining that model.

Those are very good questions. I think it's pretty much using AI, AI is going to be in the conversation, one of the things that fast-tracks it. It's also about more access to financial data, and more access to alternative data, not just financial data. If you look at the evolution of open banking, that's where the market is going. You have open banking, but then you go to places like Brazil, where they have open finance, they now have insurance in there. And now, even in the UK, I think they're going to open economy. In Singapore it's open economy, where the economy is being digitalised and you get more access to products and all kinds of data. I think that's where we ultimately have to go, using telco data, insurance data, even your social media data. I've read use cases where people look at LinkedIn, the kind of people you have on it, how many are C-level, what kind of people are in your network, and use that to risk-assess you. So I think, as we get more access to data, we're able to do more, risk-assess people, and give them more access to credit.

So, even in the UK, if you look at the payments architecture, direct debits aren't offered by the banks, they're offered by one of the UK payment bodies. I have an idea of the name but I don't want to put the wrong company out there. But direct debit is pretty much how payments are done there. We don't have that body that regulates direct debit, and the banks haven't been able to find a solution for it. Before the likes of Paystack and Flutterwave, when you applied for a loan you had to give the lender six to eight cheques, which they manually put in, and if a cheque didn't go through, they'd lost one. Now you have aspects of direct debit, but not real direct debit. There are solutions that mimic it, but I don't think there's one that has actually solved it. I think it's one of the things open banking will definitely solve. We have GSI, but that's for recovery, not for direct debit. The issue is standardisation, if everybody comes to one ecosystem, all the banks have one API, there's one standard. Right now there's no standardisation, so how can you have direct debit? The banks all have different formats, and they don't talk to each other; they go through NIBSS. I know NIBSS has a direct debit product, but I'm not sure all the banks have bought into it, because they all have different core banking systems, some have legacy systems, like First Bank, and then you have the newer banks like Providus and Wema, which are more flexible and scalable. So until we get that, I don't think there's a perfect direct debit product out there right now, but with open banking, I think it's going to be solved.

Yes and no. Initially we wanted to expand to Ghana and Kenya, and we'd applied for licences, but right now we're more focused on the Nigerian space, for a single reason: there's huge demand for our product here. We haven't captured the market; there's a huge opportunity. As I mentioned, there's the opportunity for virtual assets, and the central bank has indicated a willingness to implement open banking, which I think will happen in the next six to eight months. That means opportunity for us, and being at the forefront, we can harness it. I don't think it's the right time, when we're just about to reap the benefit of all the work we've done, to start all over again in Ghana and Kenya. There are so many lessons learned that we can take from Nigeria and turn into a plug-and-play. So for the next 12 to 18 months, I don't see us expanding that quickly, because there's so much we're learning here. We're at the perfect interval: the implementation of open banking, and then virtual assets, which everyone's rushing to regulate and understand, not just the Nigerian market. The Central Bank is doing a regulatory sandbox on virtual assets; I saw Ghana is doing the same. So everyone's rushing. I think this is the perfect time for us to understand that market, build our platform, get all the data, and then take it and replicate it in Ghana and Kenya. I think quality is better than speed. There's huge opportunity in Nigeria, the market is growing, there's investor confidence, people are getting more access to credit, and the government has a one-trillion-dollar economy goal and is actively pushing credit. So we're at the right place at the right time. We want to focus on Nigeria for the next year or so, and then try to replicate. The goal is to be pan-African, that hasn't changed, but the timeline for going pan-African has.

It's a very difficult decision, especially because my investors don't like it. Investors love to hear you're expanding in 12 to 18 months, you're going to Ghana, you're going to Kenya, you're going to be Africa-focused. But people forget one thing: the biggest challenge of scaling up is quality assurance. Ian does this podcast very well by himself. But if you now have to do it in Australia, and in 20 other countries, that means flying to 20 other countries, and you'll realise you can't. So you get other people to do the smaller ones for you, which means they're not going to do it as well as you. And even if you did fly to 20 countries, there's time and effort involved; you get tired. So the first thing is quality assurance: are we going to have the same level of quality we put into Nigeria in Ghana and Kenya? The second is time and resources. You have to be on the ground, you can't build a startup from outside. That means dedicating the same time I'm spending in Nigeria to Ghana and Kenya. Right now, we're not willing to just put boots on the ground and not be there to drive the product the way we want to build it. What's helped us in Nigeria is understanding the market, being on the ground, harnessing the opportunities. If we expand too quickly, we'll miss those. But once this thing is coasting, once we have a car that can drive itself, a team like Arsenal, then no matter what, you have a team that works: you slot in a striker and he gets 20 goals every season. That's when you can expand. Not before, when you still have 11 different players on the field who could change, if my analogy works.

No, I think we've touched on a lot. Thank you for having me, I really enjoyed the conversation.

All right, thank you, Ian.