Transformative change in credit scoring, with Sanjay Uppal

“AI is not something that's come around today, right? What has changed is our ability to store an enormous amount of data economically; the processing speeds; and the speed of transmission of information. Those three in combination are the most fertile ground to bring AI to life.”

As we’ve heard on this show before, AI as such is not a new tool, but it is now really coming of age. And today’s guest, Sanjay Uppal, co-founder and CEO at Finbots.ai, saw an opportunity to use it to transform and democratise credit scoring.

“I just said, ‘the solution has to be transformative. So we're not talking about taking a task that takes six months and do it in four months, we're talking about, 'can I bring six months down to 60 minutes? Can I bring it down to six minutes?'“

You can read more about FinBots on their homepage (https://finbots.ai/) and, as Sanjay mentioned, they aim to get back to every contact within 24 hours so if you’re wanting to know more, feel free to start a conversation right at https://finbots.ai/contact/

And due to their great content, whether you need scoring transformation or not, you should follow them on LinkedIn, too: https://www.linkedin.com/company/finbots-ai-pte-ltd/ There, you can also find and follow Sanjay or previous guest of this show from all the way back in episode 10 and now current Chief Revenue Officer at Finbots, Andre Marques

You can learn more about myself, Brendan le Grange, on my LinkedIn page (feel free to connect), my action-adventure novels are on Amazon, some versions even for free, and my work with ConfirmU and our gamified psychometric scores is at https://confirmu.com/ and on episode 24 of this very show https://www.howtolendmoneytostrangers.show/episodes/episode-24

If you have any feedback, 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.

Regards,

Brendan

The full written transcript, with timestamps, is below:

Sanjay Uppal 0:00

Traditionally, the time to value is really long, it takes 6 to 12 months to deploy your solution and then to get it working and creating value. So I just said, "the solution has to be transformative. So we're not talking about taking a task that takes six months and do it in four months, we're talking about, 'can I bring six months down to 60 minutes? Can I bring it down to six minutes?'

Brendan, when I talk to financial institutions, or other people and players in the market, they said, 'who do you think are your earliest adopters of your kind of solution? And to be honest, they are the small limit institutions, because the leap from where they are today to where they can get to with our product, is the highest for them.

The biggest guys are the slower ones, because there are cultural and other issues that need to be overcome before they adopt something that's so transformational. But you know, that's something we've typically seen in financial services.

Brendan Le Grange 0:51

Welcome to How to Lend Money to Strangers with Brendan le Grange. I've worked with - or at least near - credit scorecards for over 20 years, and on three continents. And while any actual scorecard builder among our mutual acquaintances will point out that it's been a while since I've got my fingers dirty, there is one truth I thought I still knew: that a scorecard project takes time.

Sure, not as much time as it did in the old old days, but it's not so long since I left a major bank where you'd be looking at 6 to 12 months, once you had accounted for all the committee oversight, the data gathering, the data cleaning, the data processing, the model testing, the production coding, the production testing, and only then to finally go live...

Sanjay Uppal, CEO and co-founder of Finbots.AI, welcome to the show.

We're mainly here to talk about disrupting the scorecard building paradigm but first, let's start at the beginning of your journey: we sometimes hear the line that we need industry outsiders to really shake things up, in fact, I think I've probably said it myself on here, but you were a career banker before you founded Finbots.

Albeit one who was never shy of taking on challenges in new markets. You were in India, Indonesia, Singapore, Taiwan, the Philippines, the UAE, and Malaysia. So tell me, what was your career background? And how did that shape you into the entrepreneur that you are today?

Sanjay Uppal 2:30

Thanks Brendan, in terms of my background, you know, by education, I'm an engineer - electrical engineering and masters of physics. And I think my love for technology started there. And I got an opportunity to do everything from programming chips using micro programming languages, and right up to working on mainframes.

After which I decided to pursue my MBA, which is where I fell in love with the world of finance.

And I must admit, I've had an absolutely blessed banking career gave me enormous opportunities. Before according to banking, I'd already made a few leaps of faith. But what it allowed me to really do is get deep into different parts of banking, being able to understand how different aspects of banks work, but also how do all these areas connect and interact with each other.

Not just go and try and look at just the symptoms, but really try and analyse them a lot more deeper and be able to solve problems in a lot more sustainable manner.

And yes, I've been a bit of an adventurous spirit and was never shy of saying no to going into a new market. In fact, in those days, I 'ah, absolutely, I'll head down and undertake a project or two in Taiwan', and I didn't know how far it was from Singapore at that time! So I had to get into the map and figure it out.

I finally moved in there realised I couldn't buy a car because all road signs were in Mandarin. And I didn't read Mandarin. So I was like, I'm not going to take these chances. So I spent my entire time like, you know, using taxis, and gave me an opportunity to learn Mandarin. But I think what it actually gave me was to bring me the perspective of technology, finance, and understanding the business and the external market across different environments, with banks at different stages of evolution, dealing with different challenges. And I think all of that diversity became enormously rich experiences, I look back.

And after I'd done this a few times, I saw an opportunity of how do I bring this to help a wider set of institutions than just the one that I'm working for. So the first thing that I did in 2012, I actually exited banking and set up my own consulting firm, StraitsBridge, out of Singapore, very quickly followed up with an office in Dubai, and we set up our backend in India. At our peak, we were more than 170 people.

And what it actually allowed me to do was build upon the banks that I had worked with, but gave me insights into a lot more other markets and financial institutions. And the idea of Finbots came from there.

If you look under the hood in these financial institutions, a lot of the legacy processes are pretty much done the same way they were done 20-30 years ago. And it seems surprising, but that is the truth.

And I thought that solving these problems in a lot wider sustainable fashion, needed a deep understanding, not just of technology and the capabilities we have, but also the domain, you know, how can I create a FinTech that's going to create solutions to solve some of these legacy problems. And that brought about the birth of FinBots.

Brendan Le Grange 5:30

You had had a career with leaps of faith involved in it, but after 20 years in banking, it can be difficult perhaps to take that risk when when the alternative is a nice career laid out in front of you. So you saw the gap. But still, you know, it must have been a big decision to then, from that point, say, 'okay, but I'm going to go out on my own'.

Was that entrepreneurship always something in you, or what made you take that leap?

Sanjay Uppal 5:56

Brendan, very good question. And I keep getting asked that right through since I gave up my banking job 12 years ago.

Now, I think there were a few things. One, yes, there's a bit of an entrepreneurial streak I've had. In fact, I used to earn pocket money in my college days through photography - literally taking photographs and selling them. And the reason it was a lot easier then, was most people didn't own good cameras, then. And that made my life a lot easier.

And I think the kind of roles that I took on the leap of faith for me at each stage as an entrepreneurial journey, because you're starting from scratch, you've got the classic statement of 'I'll build my parachute on the way down'. And, you know, I think having had done that, even inside the corporate world gave me that confidence.

The switch, I would say that was most notable for me was the trappings of sea level role in financial institutions with all its profile and owner office to coming back to working off a single desk as a team of one right. But I think I was willing to do that. It was something I realised that there was no other way to do this. You guys, I tell people yet, when you get on to the entrepreneurial journey, you are never prepared enough. And I think having prepared myself mentally helped, but there were a lot of discoveries along the way.

But what I really appreciated through the journey was the goodwill of a whole host of people I met during my career, who became our references. So I ran a consulting firm for 10 years. And we never advertised. We never sent out email flyers. We never cold called. Every client was a referral or repeat business or somebody, me or people in my team, had in their network, right. So I think we were fortunate in that sense. And it really was an amazing journey. And I was really fortunate to have a lot of really brilliant people join me along that.

Brendan Le Grange 7:53

Yeah, you echoing some of the things that Jinesh Vohra spoke to me about. After 14 years in Goldman Sachs, he started his business and he said he suddenly realised that in Goldman, when he needed a question answered about the law, there was a team of lawyers, when he had a question about marketing, there was a team of marketers, and suddenly, he was on his own working it all out.

But I think what's interesting in your entrepreneurial journey with invites is how it's evolved and grown through that whole journey from bootstrap to Series A. So before we start talking about the product itself and what you're doing, you know, what's that journey been like? How have you seen Finbots grow?

Sanjay Uppal 8:31

When I started, I actually wrote down three key principles of what we want to do in our solutions for financial institutions. The first one was the solution has to be transformative. So we're not talking about taking a task that takes six months and doing it and four months, we're talking about, 'can I bring six months down to 60 minutes, can I bring it down to six minutes and do it even better?'

The second thing was around the democratisation. When you look at high complexity areas in financial services, the target market for those b2b solutions is typically top five to 10% of financial institutions globally. The rest of the organisations either can't afford those solutions, or even if they can, can do not have access to the skills that are needed to drive the solution. So I needed products that are going to be democratised in terms of affordability, the largest global banks down to the smallest micro lender, but equally have a design that would allow both of them to use it effectively.

And you know, that came to me at a coffee I was having with a friend of mine and we haven't got coffee, and I was looking at something on some FX rates because I had to send some money. And I'm looking at the app and the guy who was serving us coffee says I use the same app. And I just put the phone down or spoke to this mate of mine. He's also a banker. And I said 'We spent a lifetime of our careers where if I had to remit money in foreign exchange to another country, I had to go to the branch, queue up, fill up this complex form, go through that type of process and hope the money would reach the other end in three, four days. Today, whether it's me or this guy who just served me coffee, we are able to remit this money literally on the fly within five minutes. That's democratisation. Why can't other solutions we like this why contact create a b2b solution that even the smallest FinTech lender who's just started or a micro lender is able to use?

And the third thing for me on a B2B solutions, where traditionally the time to value is really long, it takes 6 to 12 months to deploy your solution. And then to get it working and creating value.

So I just said, It's 12 months, we bring it to 12 weeks. And you know, we really fortunate that we're able to achieve that today.

So time to value for our clients is 12 weeks, the solution goes in there, they 're able to create the models and roll them out. So the idea essentially, was transformational product, democratise access to scoring tech, and rapid time to value.

That's where the journey started. And we actually developed the first version of a product and rolled it out in February 2020. And it was enormous ly successful. I think the quality of scorecards our product was developing the impact it had on a client, that solution is amazing. So we were okay, we ready to take this to a bigger scale now. And March everything shut down, right.

So this is 2020 and COVID struck, we were sitting with a proven successful solution, we were ready to take to a wider market and suddenly, we couldn't do that anymore. We then spent nearly the next I would say 18-20 months on regrouping ourselves as an organisation. But the second thing I did was, I looked at the entire product roadmap ahead for the next three years and said, 'we've got to do it as soon as possible'.

We don't have an opportunity to go and sell to customers, the customers weren't in the office themselves. And we just use that entire time to hunker down and speed up our development speed up a lot of our experimentation. And in a way, it proved to be a good time because we didn't have distractions.

I also decided to bootstrap it. So while the funding market was doing pretty well, the thought that I had is I have set aside money that I want to invest in this venture. And I want to take to the market once I've proven our idea. So the first time we reached out into the market was December last year - so literally a year ago.

When we started speaking with investors, we had a product that was ready customers, and validation from external partners. PwC was already pitching our product to their clients and had evaluated it and were very satisfied, we had been selected to global accelerators amongst literally a handful of fintechs. And with that, behind us, when we reached out into the investor world, there was some fantastic feedback. Obviously not everybody had a fantastic feedback, some have said, well, this is not a sector we look at or this is not what we do. But I think the guys who were interested was really good.

And we were really fortunate because Accel moved very quickly. And literally three weeks later, we had a firm offer for them to take up the entire round.

Brendan Le Grange 13:01

Yeah, and I think not just investor confidence, I've seen you awarded among the top 10 fintechs in Singapore by the MAS there. So getting good response.

And I think that clear sort of three prong approach is probably behind that, to really be able to reference some guiding styles that are there clear, in your mind definitely makes it easy, I think for the team to work in the same direction. And yeah, I think we'll get into some of that stuff around turnaround times and things which people can really appreciate.

But one I hadn't thought of was that democratisation. And I've worked in in credit risk for for 20 years, and it has been either you're a big bank buying from the same global name that everybody in the US and the UK and it sort of the big developed markets is this sort of handful of names, we could probably list off the top of our heads now that sell those solutions. Or you're buying from somebody down the road who's running a two person consulting firm out of their garage, who knows how to build scorecards. And there was very little in-between - and it had not struck me as unusual!

But you're right. There's no reason they should be that massive gap, that empty part of the market.

As you said the big focus also on turnaround time, and yeah, I was just talking to Frank Gerhard from McKinsey. And yeah, he was saying, 'if it takes you six months, and it's sometimes even longer than that, but if yeah, if it's taking you six months to get a decision model into the system, it's just not going to work in this current environment. We need faster turnaround times.

Obviously, we've all seen the impact of COVID. And we've gotten into now inflationary pressures. It's not a world anymore that we can rely on, you know, this spend a year building a framework and it's going to reference two years of stable data to predict the next two years or stable outcomes. So really underlying that need for turnaround time that you've highlighted as a key part of the Finbots philosophy, but within that there's obviously a sort of a conservative alarm bell saying like, well, two parts I guess, it's (1), have you been able to leverage sort of modern technology to do that? And (2) in terms of the modelling and the performance of the controls, all those sort of checks that a banker that you're trying to sell to might want to look into?

How do you get that same level of statistical rigour, that same level of control, but in much faster times than we were used to just a few years ago?

Sanjay Uppal 15:25

Very, very valid points.

And, you know, that was literally right at the heart of the product design itself. At every stage of our product, one of the thoughts I always do is flip my hands and say, 'hang on, would I buy this? Am I happy with this?'

So you have to think about this was three, four years ago, when we were doing the core design, and we're deepening the product capability we started on, and I was already building into it, the principles of fairness, explainability, and transparency. With us putting our banking hat on, I don't want something that's a black box. Right. So that was the core of our design.

If you have to remember what we talked machine learning AI today is not something that's come around today, right? What has changed today is our ability to store enormous amount of data economically. Number two is the processing speeds we have today. You know, you want a search bar before you type, your third word is already telling you what it should be. So think about it. And there are millions of people doing it at the same time, any second. And the third thing is the speed of transmission of information.

I think those three in combination literally are the most fertile ground to bring AI to life.

And that's what we've essentially done. But be mindful that when you're doing things at that speed, there are things that could happen which go out of your control.

And, you know, I don't know whether it's a fair comparison. But if you look at, you know, nuclear energy, or nuclear energy harnessed, right, as the most amazing thing, most efficient in terms of producing power, energy, etc. But when you lose control, you can have a Chernobyl. And I think when we've handled AI, we've literally handled like that, in our product development and design. minimising what are called trust me, it's going to work well factor, because I think anybody who's gone through a few crisis and financial institutions will realise that what always let it on was with somebody said, trust me,

Brendan Le Grange 17:18

And now we've also got leadership that understands. So now we've got people like yourself, people in senior roles, who've built the careers up as analysts, or done lots of on-the-grounds, hands dirty work in the space of data in general, and even in some of these technologies.

Sanjay, I guess one thing I want to get my head around, and maybe some of the people listening would be wondering as well is where do you get the data that feeds these models, or what's the data process for the Finbots solution.

Sanjay Uppal 17:47

The data side remains very core, it's at the heart of what feeds into developing your models. One of the things we've been mindful about as the perennial challenge of lack of data - there is an enormous amount of data organisations already process data that's readily accessible and it may or may not be a bureau data. So we've made our product, data agnostic.

And what it means is that you're able to bring in various kinds of data, numerical, categorical data from different kinds of sources, and be able to effectively use it in the model development. So we've got a B2B lending client who's got no access to, or they are not even credible annual financial accounts of the entities they're lending to, they don't have access to bank records. And he said, even that is not really relevant, because a lot of transactions are happening in cash. What they do have is credible data on the entire supply chain, or the purchase behaviour on each of these entities going back to I think about a year or plus, and that is verified data.

For me, honestly speaking, that data is more reflective of the underlying business activity of the institution you're looking to learn to then a point in time annual accounts, which will get produced once a year, or once a quarter or bank statement, which might be at a summary level.

And using that data, they're able to produce their scorecards amazingly well. But equally, there's always a bit of caution, which is just don't throw all data because there's this term of big data, I think you're still looking at good data, you know, you're looking at relevant data. So I don't think AI has reached that point where no matter what you throw in, you will get a magical outcome and that the key thing still remains that there is a human judgement and war but the system does an amazing whole host of things but don't take your eye off that was you know, if you look at AI, we've seen some spectacular failures, right even from some of the biggest names globally.

And that's because I think that care not taken in terms of how you using the data, what you're bringing in to train your models. And you know, when I talked about fairness in the model, that's a key piece.

Brendan Le Grange 20:06

Yeah, because I think that's actually one of the unmet promises of AI and machine learning in many situations, that we were promise 'big data, you know, just throw everything at the machine, and we'll give you an answer'. And very few financial institutions had that sort of data; or when they had it, knew what to do with it; or when they knew what to do with it could turn it into a scorecard. Maybe it's because of your consulting background as well, but you've almost built in that consulting ability of how to use the data. So it closes that loop, which I think is exciting, because it's in many ways, doing two or three steps at once. Yeah, it's always exciting to see real AI real machine learning at scale.

One of the things, though, when you talk about your sort of transformation of the industry, as you said, you didn't just want to go from sort of 12 months to nine months, you did 12 months to 12 weeks, or six months to six minutes. It's not just about efficiencies, it's not just okay, we're going to save my client a few thousand dollars in terms of how they roll out credit scorecards, or even I'm going to get them a few extra gini points in a model, and that's going to save them some money or make them some money in their lending.

This is like a it's a reimagining of their businesses, its proper transformation. What have you seen it do for lenders? And how are you seeing your customers react and respond to this new ability to really rapidly implement models.

Sanjay Uppal 21:23

Today, we have a client in Nigeria, who's using it very successfully. It's an IFC BAP FinTech lender, and it's just been amazing in terms of what is shown to their business there. Enormously successful.

As we speak, we are about to go live with the first neobank in Mongolia. So apart from the traditional markets that you and I know, and you know, you would expect the product to go live, what we're doing is also going to a lot more other markets to prove that this capability can be used there, and can empower our clients there. So we continue on that.

Oh, by the way, we just signed earlier today, our first client in Australia. And the idea essentially is the product is equally suitable for a developed market. Somebody asked me recently, what's the biggest challenge in your taking your product to market, right, with the clients you speak with. And my answer: master disbelievability.

Because I've heard senior credit and risk guys tell me, 'look, this is how we've done credit models my entire career, I started 30 years ago, this is how we do it, this is how we do it today'.

So we actually set up live demo environments, which means we've told you about the product, now see it. Because I think there are design aspects in our product that are very unique. And the term you used was right, which was 'reimagined' - we have reimagined the entire process. But when I tell people about our product, their friend to fit what I'm telling them in the legacy framework. So we usually very quickly cut to the product, so they get to see what it is. And it pretty much every time including literally the reason I was late for this call was it leads to a ha moment. So we had one of the largest global banks who started sceptically. And we close by saying, okay, guys, we are running out of time, but we need 30 minutes more, can we continue tomorrow.

And this is the global heads and stuff like that.

So I think seeing the product really makes a difference. So last year, we started ensuring we do that with anybody we are pitching the product to what we have now done is Brendan, this year, we set up I think about 15 sandbox environments.

We can take organisations through end to end, two week POCs, where they bring their own data, develop their own models, and they get access to the platform, so tested out and literally test drive it - literally we can start a POC tomorrow if somebody signs up today and says I want a POC. Now they're seeing it with their own information data with their own hands. They're using it doing whatever needs to be done. You know that that's helped us move on.

The second thing, obviously, and you touched upon it earlier, the the market environment is seeing its first real credit, I would say 'challenge' emerging across the world for the first time in nearly 15 years. So there is also a demand side where people are saying Helen's, it sounds good to be true. But let me have a look at it. And you know, that's given us a lot more traction now.

Brendan Le Grange 24:10

Yeah, and I think it is a we talked about 15 years since the last crisis. And the last time that maybe Boards at the banks have really been interested in the space. And it's also probably since when last we saw big revolutionary changes in terms of the scorecard technology. I mean, of course, there's been developments and improvements in modelling. But you're right, a lot of the more innovative use of different techniques and technologies was on the fringes may be used in some marketing capacities but wasn't really pushed because the models were working, the models were signed off. And nobody really wanted to shake the boat when it came to that and there wasn't an obvious need to so in many cases, it was easier if you were a career risk person to tweak them to make them a few percent better every year, but maybe not a high priority. were suddenly yeah, as you said, you you use that period of COVID to to buckle down and work and distracted, as we emerged from that, you know, into the promise of a roaring 20s That never never emerged. And suddenly you were there with the right solution.

And I think it's, yeah, clearly, as I said, from throughout your career, you've taken on new challenges, you've you've obviously got a big focus on on growing and trying new things. Before I let you go, I'd be quite interested in in kind of where you where you're looking next where your energy is going.

Sanjay Uppal 25:16

And Brendan, I think we're kind of working in a field battle streams, we have a product roadmap and a journey that kind of continues on our development side. So there's an organisation, part of the organisation that's focused on developing newer capabilities and versions within our solutions, we are adding more functionalities. And that's going to continue over the next couple of years, we have a roadmap for that.

The second one, as you rightly said, is taking our product to the market and ensuring every client is successful. And, you know, it's meeting their objectives. And you know, that part of the day to day running continues. At the core of it, you know, and I always present the numbers, which I you know, to our clients, which is the scale of financial exclusion we have globally, whether it's lending to small businesses or consumers.

It is amazing that even after all these years of financial services, and this came out in IFC report last year, the scale of financial exclusion is multiples the size of the actual market who has access to credit today, especially in developing markets. So one of the things we look and work with our clients is, what are they being able to enable in driving financial inclusion, if we which could be just having more fair scorecards, which means the word of traditional biases, but not to a lot more accurate credit assessment, which allows them to confidently lend to the excluded borrowers.

And finally, the third thing is, I think, as this progresses, we should expect last rates to improve which means that the good borrowers who have never defaulted on not being charged a premium that will make up for the bad barbers, because it's not well, we've got a great product, we sell it to somebody and we move on. But, you know, we recognise there's an opportunity that they have, and there are people and businesses that should benefit from this. So we actually work with our clients to look at that aspect, too. So I think that part of the journey will continue, even as we are developing and harnessing more capabilities and building them into our product.

Brendan Le Grange 27:31

Yeah, I think that's one of the really big impacts is that a lot of the inefficiencies in the credit process at the moment are paid for by the borrowers who end up getting the loan. So if it's hard for somebody to measure risk accurately, the prices are higher than they need to be, because people still need to borrow money. So some people excluded other people are included, but they pay a lot more. Getting rid of those inefficiencies, sure it makes banks more profitable and banks more stable, which is good for the investors and the cost of the capital and such, but it's the consumers have benefited a lot because more people get access to credit and those get access are cheaper.

And I love the global reach, you've gotten jealous of some of the markets. On the show, I try and get guests from all over the world. And I think I've had 40 countries represented and yet to get somebody on from Australia, so I'll have to have to take some tips from you. But I think that also means you're talking to people talking to clients or able to talk to clients from as far afield as Mongolia to to Australia, I think people listening today, that means no matter what market they in, if they're interested in what they've heard today on the fin bot story, maybe they want to talk to you to set up one of those immediate proofs of concept, where where could they go.

Sanjay Uppal 28:45

I think the easiest way would be because people are sitting across time horizons is just go to our website and drop in our details to the Contact Us page, and we respond within 24 hours. So it's not something that goes in there, and you're hoping you'll get a response, you will get one.

The second piece is obviously following us on our LinkedIn page. And that's where we are posting some progress very often about and some of our thought processes around, you know, what we've seen in the market. You know, how some of the success stories have proven for our clients, and especially in the current environment? So yeah, LinkedIn is a great place to follow us to

Brendan Le Grange 29:20

Great yeah, we'll put those links in the in the show notes as well. I've been following the page and as you said, some really nice mix of content that comes through in terms of things like the announcements of the same bank deal, but also stuff to read and to keep a track of what's happening in that space. So certainly recommend people click on there and and follow you guys.

Sanjay, thank you so much for your time today. I mean, it's been paradigm-busting for me and really interesting to challenge some of those assumptions that build up over a long career in her space and to really think here, why? Why were we putting up with that?

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.