1013. News: Big moves in the mortgage market - and is everyone using AI to manage their finances?

0:04This is Fintech Insider News. This week, lending rules loosen as HSBC offers mortgages at six and a half times salary. US open banking on hold as a judge temporarily blocks the CFPB's rule. And over half of British adults are now using AI to manage their money. We take a closer look. We'll be discussing all of this and more on today's show. So don't go anywhere.

0:42Hello and welcome to episode 1013 of Fintech Insider, brought to you by 11FS, the five-time consultancy of the year that works with banks, digital banks, investment firms, and fintechs, building the next generation of financial services. I'm Benjamin Ensort, Director of Research and Strategy here at 11FS. This week, we celebrated Bonfire Night here in the UK. And as the tradition goes, fireworks lit up the skies across the country. Another area bursting with activity this week is the world of fintech, with many stories igniting widespread interest. Helping me sift through the sparks and the smoke is our panel of fintech experts.

1:20So let's meet them. First of all, a very warm fintech insider welcome to Jermain Bari, co-founder and chief growth officer of Zazu. Welcome to the show. So now I gather Zazu is at a very exciting time. Can you tell our listeners a little bit more about it ahead of the launch, please? Yeah, sure. Thank you for the invite. It's a privilege to be here with you guys. Yeah, a little bit about Zazu. So we are building and launching very soon a business banking stack, basically for African entrepreneurs and SMEs. Probably the closest model that you know in the UK is Tide, or the Mercury of Africa, as we've been called, where we basically helped entrepreneurs simplify their everyday banking, right, by being able to pay and get paid, but also manage things like bookkeeping, invoice management, spend management, everything in one place.

2:16And yes, so we are looking at two markets at the moment. The first one is South Africa, where we are live with a small pilot, and in Morocco, where we will be launching in the coming weeks and with a bigger and larger announcement probably around December. So launching at both ends of the continent. Fantastic. We also have a FinTech Insider debut for Ian Armstrong, Executive Director at Comply Advantage. Welcome to the show, Ian. Can you tell our listeners a little bit more about you and a little bit more about your role at Comply Advantage, please? Thanks so much, Benjamin. We're really, really pleased to be here.

2:56I'm a long-time listener, you could say. Yeah, so my background before I joined Comply Advantage really is in financial services for getting on for 25 years. I was originally in the enforcement division of the regulator here in the UK. And after just under a decade of that, I moved into banking. So I was at Barclays, followed by HSBC, followed by NatWest, where I ended up as the head of risk of metal and NatWest Boxed before NatWest Boxed had a name. So I've been at Comply Advantage for just around three years. And my role at the company is really kind of cross-functional role that cuts through commercial product and marketing.

3:45and really I'm trying to use my background in the industry to help effectively any member of staff in the company to understand the needs of our users, so the needs of compliance teams, risk managers, etc. We love a bit of customer needs focus, so fantastic. Welcome. And I'm also delighted to welcome my colleague Rosie Lee, a Senior User Experience Researcher and Customer Strategist here at 11FS. It's great to have you with us. Could you introduce yourself to our listeners, please? It's great to be here. Thank you for having me. Yeah, for those listening at home and for our guests, I'm Rosie. I'm a Senior Customer Strategist at 11FS.

4:30So really, I just spend my days talking to customers, understanding what they need, taking those insights to banks and financial institutions, helping them to improve their products and services, and ultimately improve customer experiences. And I'll very much be bringing that perspective to our discussions today. Fantastic. Welcome. Okay, well, let's get into the news. So we're going to start with a story from the UK that may bring a little bit of deja vu to some of our listeners, which is that HSBC is offering six and a half times salary mortgages as lending rules begin to loosen. So HSBC is now offering its premier clients mortgages of up to six and a half times their salary with at least a 10 % deposit.

5:13the highest ratio from a British lender since the 2008 financial crisis. The move coincides with a review of the lending rules that were brought in to prevent a repeat of the risky borrowing that preceded the 2008 financial crisis, encouraged by the British government, which is trying to help more people get onto the housing ladder. So this shift also signals growing competition among lenders to try and stretch income multiples amid pressure from the government to stimulate the housing market. Separately, another UK lender, Nationwide Building Society, has announced on Monday that it's expanding its range of interest-only mortgages.

5:53Meanwhile, over in the United States, Robinhood Gold subscribers are now able to access discounted mortgage rates and closing credits through a partnership with digital lender Sage Home Loans. So, there's a lot to unpack here. first of all perhaps maybe come to you Rosie first how difficult do you think it is for people to get a mortgage is that a difficult process is this justified lending people more money does that help I mean we've in the past we've we've lent customers more money than perhaps they are being lent right now I do think it's hard to get a mortgage I myself have been trying I get one recently.

6:39And I think that this article, unfortunately, is a bit headline grabbing. It only applies to a small group of customers, high income borrowers. They're those premier customers that are earning 100 ,000 a year or they have the same amount in savings or investments. So it's not really expanding access to mortgages across the board for everyone. It's only providing more flexibility to that more affluent segment. So you're saying I'm being sensationalist? Possibly. Ian, as someone who's worked at the regulator, I imagine you saw sometimes pressure from governments that are trying to achieve economic objectives.

7:25And yet at the same time, the regulators are trying to make sure that the financial system is robust. When I saw this story, it was like, wasn't, you know, wasn't sort of generous lending, let's call it that. One of the root causes of the financial crisis, or am I being too conservative? No, I think you probably received the headline exactly as it was meant to be received. I'm kind of in Rosie's camp on this one. I remember the 2008 period, you know, the global financial crash very well. I was actually at the regulator when it happened, not to mention the years immediately after it. I remember it quite vividly.

8:06And I think one thing to remember is that prior to that, there were a lot of people, even in this country, who were getting self-cert mortgages. So in other words, they were just certifying themselves as being able to pay whatever cost was involved. So I think, again, I'm very much with Rosie on this. I think as long as there's robust processes for banks to understand exactly what the income part in the loan to income equation consists of, I think there's a way to offer these mortgages safely. And it sounds like it is a select group who's being given access to them. But yeah, I think the headline clearly is designed to kind of evoke memories of the 2008 crisis.

8:52I think if, sorry, when the next financial crash comes, I think it's much more likely to be driven by, say, the commercial property sector rather than retail or even the private debt industry. And in terms of whether it helps people, yeah, I'm not too certain. I think it could potentially help people feel that they have access to a bigger pool of potential properties, right? You know, because if you're looking to buy a home, you might say, well, I can, you know, realistically afford these 10. With this change from HBC, maybe that gives you access to another five if the bank's given them assurances that they're eligible for that level of lending.

9:38So it might help to, you know, stimulate the growth agenda, to your point. Shemay, I'd love to bring you in as well. What do you think of this? Do you think we should be welcoming a sort of loosening of lending standards and sort of more people being able to sort of borrow more? Or should we be a little bit nervous if lending standards get loosened too far? Look, just as a disclaimer, I'm not the mortgage specialist, but looking at both continents, right, Europe and now Africa, where I'm spending most of my time, we definitely see a lot of the same processes, right? That are still very lengthy.

10:22There are still very much paper-based, even more so here. So that brings, as Jan said, lots of problematics around access, particularly around, you know, self-employees, the gig economies. A lot of actually of the, to a certain extent, some of the business owners as well that will be serving. So look, I think there's definitely an opportunity here. I think that's where, you know, probably some of these headlines now are on point. It's probably the next wave of innovation, especially when it comes to data and alternative source of data. We're starting to see it here as well in South Africa, where you're seeing more and more marketplaces.

11:02You're seeing more intelligent tools to serve that. So I do think there is a need for, you know, reviewing some of these processes and a better sort of assessment in scoring models, which were basically traditional models, are still not yet a fit. You make some really, really interesting points there. I mean, Rosie, I'd love you to sort of reflect on that a little bit because I'm sort of looking at HSBC and thinking, well, are they offering to lend more to sort of try and compete, sort of try and make themselves more attractive? But actually, as borrowers, as customers, people trying to buy houses and so on, Are people looking to borrow more or are they looking for an easier process?

11:42Are they looking for both of those things? What do borrowers want from the mortgage market, do you think? Well, they're really looking to kind of live in a home at the end of it, aren't they? So I think the concerns come around, yeah, creating a bit of vulnerability in borrowers. if they're borrowing so much more than their income, it helps them to get on the ladder, but then are they less safe if interest rates rise, incomes drop, living costs increase, they've got less room to manoeuvre. Nationwide's doing a similar thing with their helping hand scheme, which is allowing first-time buyers to borrow up to six times their income, but only if they have a solid credit profile.

12:30So I think it's part of a broader pattern, broader theme from banks and they're experimenting at the edge of that regulation. But they're only prioritizing the customers that are at the least risk. So both HSBC and Nationwide are sort of doing the same thing of saying for certain customers in certain situations, we'll lend you a bit more. but I suppose Ian coming back to your point about self-certified mortgages that feels quite different to the self-certified mortgages where I say I'm earning a huge salary so lend me lots of money even if I wasn't. Yeah absolutely I do think you know we sort of live in different times now definitely take the point that there is there is always that risk if you kind of increase the amount you're willing to lend to a person that they are going to default on that.

13:23I do definitely think there are better controls in place around that kind of thing right now. And probably what we'll hear, you know, in a few months time is stories about people who were not able to access these mortgages. They were turned down for them. That would not surprise me at all. I suppose all of us who lived through the global financial crisis have a kind of sort of instant reaction to anything that sort of smacks a little bit of a return to some of the sort of the worst practices that sort of help to trigger the financial crisis. Okay, let's move on to our next story, which is that ComplyAdvantage has launched an AI-native platform to combat surging financial crime.

14:05So ComplyAdvantage has launched Mesh, which is an AI-native system that brings together customer and company screening, transaction monitoring, risk scoring, and real-time payments analysis. Built on large language models and predictive machine learning, Mesh uses agentic AI that can learn, act, and adapt across the full compliance lifecycle. The launch responds to reported surges in AI-enabled financial crime over the past couple of years. According to Salent, a research company, banks report anywhere between 75 % and 90 % false positive rates across their fraud and AML transaction monitoring systems.

14:45So Mesh by ComplyAdvantage aims to cut the number of false positives by up to 70 % and speed up investigations, turning compliance into a competitive advantage through faster onboarding and smarter transaction processing. Ian, so So congratulations on the launch. It's great to have you here to talk to us about it. Tell us a little bit, what does it mean to have sort of AI native in a compliance platform? I mean, compliance teams have been using AI for a while, right? How is Mesh different from what firms have been doing up till now? Yeah, it's a really fair question. And it is an important distinction because I think, honestly, there is a lot of noise out there in the market right now.

15:27with basically everyone and their uncle claiming to be AI powered. So I'll try and be as specific as possible about what we mean. So traditional anti-money laundering systems, and I've seen plenty of those in my time, were designed really in an era of batch processing, on-premise systems, rule-based logic. so when some activity or customer attribute triggered a rule it created an alert and a human investigated it. Fine. What a lot of vendors out there have done is essentially apply a layer, a bolt-on of AI on top of that type of architecture perhaps using machine learning to score alerts or to prioritize alerts in a queue.

16:19Don't get me wrong, that is helpful. But it's fundamentally limited because the underlying infrastructure wasn't really designed for it. And so you still lack an understanding of context. That's why you end up with those very high false positive rates that you referred to. It always staggers me when I hear stats like that. But I mean, they're absolutely accurate in my experience. Because people still have these data silos, are still effectively copying information from one system to another. And so the AI is trying its best, but it's essentially an add-on rather than integral to how the platform thinks, if I could use that kind of language.

17:01So what we're built with Mesh is genuinely different. It's three adaptive layers with a kind of common operating system. So at the foundation, the thing that breathes fire into the whole thing is our proprietary intelligence and data layer. So that's continuously ingesting data from out there in the world, from things like sanctions lists, from media articles, from government warnings. And then on the other side, from our clients, so transaction patterns in their customer base, customer attributes from their CDD, processes. And that first layer really is imbued with AI from the outset. So there's use of LLMs in terms of how we process documents and classify them according to what we call our risk schema.

17:57There's sophisticated machine learning involved in how we resolve entities. So that's, in other words, how you say, oh, this person here and this person here is actually the same person. So that's entity resolution that is driven by artificial intelligence. And on top of that are what we call our risk applications, our risk apps sit on that. So these are things like customer screening, transaction monitoring, payments analysis. But critically, they all draw on the same data model, the same entity resolution, same knowledge graph. So when a client is investigating a customer, they're not switching between five different systems with different views of the same person.

18:42It's one unified intelligence picture. Now, there are vendor solutions out there that try to orchestrate different systems. But the overall effect isn't the same because they're effectively trying to get, say, three or four different things onto one page. they're not really doing too much to kind of integrate them or make those four different things talk to one another. And then the third layer and the third layer with AI also imbued into it is we've introduced our first user facing AI agent, who we call Cassie. I say first because there are a lot more in the works already. And I say user facing because we have a load of agents that we use for internal processes.

19:31But what CASI does is orchestrate workflows across the entire platform. So rather than an analyst manually gathering information and cross-referencing databases, documenting findings and making decision on a case, CASI can execute that entire workflow autonomously for lower risk cases and surface the genuinely complex things that need human expertise. So in short, AI native system, the entire workflow from data ingestion through to investigation to disposition is designed around what AI can do well. And the idea is that humans are then freed up to focus on what they can do well. Amazing. Jermaine, I'd love to bring you in here.

20:21As you think about starting operations in South Africa and Morocco, you'll be doing lots of good for all sorts of businesses but also presumably there'll be a few bad actors there'll be a few bad people trying to get in and take advantage of you, take advantage of your systems people who shouldn't be there or doing things they shouldn't be doing and so on what do you think of AI platforms like Mesh are you thinking about this sort of thing to try and protect yourselves from bad actors? No, definitely and that's again even more so a topic in Africa rising on. And this is where also there's a big wave of new compliance softwares and so on.

21:01The likes of, you know, the earlier version of Compliant Vantage. I know, remember the times when RecTech was not even the word and Compliant Vantage came in. So we're seeing a new wave on the continent coming in to solve exactly these points. That is true that indeed in our early days when we started operations, we did feel that it was quite, The experience was quite fragmented to find a software specifically for identification, the ML and so on, and then everything around transaction monitoring and so on. So definitely a need. This is where also that's typically a big use case for us in the application of AI at the operation level.

21:44Obviously, there's another big use case for us, which is around, you know, specifically customer facing. but definitely something we're looking at and definitely something we want to invest more specifically as I said in region where both we know that that hackers have been very successful, very active Nigeria is one, is a market that we're looking at where in the meantime we've had a lot of very successful fintech and managed to scale quite importantly but in the meantime we're exposed to certain to certain fraud and hacks. But also you'd be surprised in some markets as well where the banks don't always welcome certain fintechs.

22:33And one of the ones that compete with them. You know which ones. But that sometimes may have some misintentions and try to even poke or hack the system. So definitely relevant more than ever. Yeah, and Germain makes a series of really interesting points there. But one of them that strikes me is the differences between countries, right? Because we know that one of the challenges with sort of know your customer, anti-money laundering, fraud, and so on, is that just different countries have different systems. And even things like the frequency of surnames, you know, family names, I mean, in some countries, it's much harder to identify an individual than in another because, you know, if you've got 20 % of the population share a family name, it's much, much harder to tell individuals apart.

23:15Are you finding, Ian, that AI is sort of helping with some of those country-specific challenges where some countries you've got some really quite good data about who people are and other countries you just haven't? Are you seeing different outcomes? No, no, I think we're somehow, we're still complaining about Europe not being fully standardized, but I think we've came quite a long way, especially when you look indeed i i use loosely africa as a continent but indeed you can separate in regions and then definitely look at country specific and within sometime countries you have you have uh you have uh areas or sub regions where where you have different uh almost jurisdictions um south africa being one but um but indeed you you you're looking at different types of level of digitization from the government different types of ideas uh morocco luckily enough is you know has been pushed by the government around a full digital uh digital ambition including uh you know uh the idea of uh of of having a um an ekyc system and digital identification so that's something we can definitely piggyback um but indeed uh that is something where as i said we we um that's one of the key criteria that we're looking at when we when we partner and we're looking at uh specific vendors is to make sure that some of them luckily enough have developed across certain regions and have developed these specificities and local flavors but it remains a challenge specifically around indeed identification and things like this and this is where also I think AI has brought up a big leap where I think a lot of whether it was false positives or specific processes were still at some point flagged and then manually manually reviewed where now we are able, and this is where we can get an edge, still in countries like Morocco and South Africa where the banks are quite developed.

25:21But if you look closer at certain internal and operational processes, where we can definitely be more efficient. Thank you. It's a really interesting point that you made, Ben, about the kind of prevalence of different names and the way names are structured and things like that playing out differently, you know, in different parts of the world. Because that is absolutely one of the kind of core contributors to the problem, if you like, of name screening and those false positives that you referenced earlier. Another thing that we've done in the course of building MASH is to look at name commonness.

26:02commonness. So it's very much trying to address the point that you've just made, which is in a particular country, let's say, or region of the world, certain types of names, certain names are just going to be more prevalent. They're going to be more common. The example I give is when I first started, the first bank I worked in, there were three people with my name in the same building. So, there was I thinking I had an incredibly unique name, but in fact, I was sort of brought crushingly down to earth when I realized that it's actually a really common name. And yet, if I were to go to, I don't know, a country in say Asia Pacific, it would be a less common name.

26:50So, our models, our kind of search models, search and match models, have been trained on that type of commonness. So they're hopefully able to detect that this name in this country is relatively common. And so when they're trying to match against, say, a sanctions listing, that's part of the calculation. When they are calculating the likelihood that this name of your customer is the same as this name of, say, a sanctioned individual. Presumably, you've been sort of testing and deploying Mesh into different countries with different clients. Are you finding any sort of differences between countries?

27:32Are you finding that it's just a bit quicker to reduce false positives in some countries than others? Or are you finding it's performing similarly? I'm just curious. Are you noticing any differences in a system in different countries? It's a great question. So we've rolled it out, I think, to 200 or 300 clients at the moment. now, I must confess, I haven't seen the exact country coverage of those clients. We have clients in our entire base in over 90 countries, I think. So we've learned a huge amount. I mean, Client Comply Advantage has been running for 11 years. So in that time, we feel we've really kind of earned our seat at the table.

28:18We've learned what works, what doesn't. And to your point, We've learned a huge amount about what works in specific markets. So I think it's possibly a little bit too early to say because we're quite, you know, I mean, we've only really recently launched it in terms of those variances. But as I say, the way the models have been trained is hopefully going to make it so that wherever you are in the world, you'll see immediate benefits from using the tool. One other quick question for you on this. Fraudsters and criminals are notorious for being increasingly international. I mean, obviously, there's sort of low level fraudsters are just attacking people in one of their own country, but there's a lot of international fraudsters.

29:05Is the system sharing information across those clients? Because let's imagine Rosie's an international criminal. She's probably attacking lots of different clients in lots of different countries. Are you sort of sharing, is the mesh system sharing data from different clients so that, you know, the second bank that Rosie attacks says, hang on a minute, we've seen her before. Is that how it works? Not at the moment, no. I mean, I would very much say watch this space on that. I think that type of data sharing is something that we are extremely interested in now that we've got this unified platform.

29:43I think it definitely opens the door to that type of module or that type of functionality because, yeah, there is a huge amount of interest in it. I think historically, it would have been quite difficult to do that, firstly. And secondly, of course, the big concern there is always around PII, so how you handle personal data, which can be fraught enough in one country, let alone doing it across borders. But I think there is a lot more appetite from both the industry and, in fact, regulators, governments, for that type of data sharing. So I would say watch this space on that. Fantastic. Well, congratulations to you and your team on the launch.

30:32Okay, we're just going to take a quick pause here and we will be back very shortly. Hey, folks. David Breer here, CEO of 11FS. Here's something you might not know about me. I get a lot of people trying to impersonate me online. Fake profiles, scam emails, the lot. And a big part of that comes from data brokers. Hundreds of them quietly collecting and selling your personal information. Your phone number, email, home address, job title. All out there and all fueling identity theft, scam calls and spam. If you've ever searched your own name online, hands up, who hasn't? You'll know how exposed you really are.