The XBank Podcast
We welcome Aubrey HB, Director of Advanced Analytics at Nationwide Building Society to discuss the uses and effects of AI and machine learning in banking — where we are in the journey and what we need to consider as we move forward. Host Fran Sorrentino is joined by Zachary Scott and Simon James from Publicis Sapient’s Financial Services team.
Nick Edwards from Lloyds Banking Group joins Matt and Fran to discuss the challenges of scaling to Enterprise-wide change.
Geraldine Maringo, Senior Product Owner at Nationwide Building Society, joins us to discuss organisational Customer Journey Transformation.
You’re listening to the XBank podcast, a series of conversations, exploring the how in digital banking transformation over the coming episodes. We’ll pick apart the concepts. Look at the practical steps and analyze live examples, alongside industry movers and shakers. If you’ve ever wondered what it takes to get transformation done. Then this is the podcast for you. I’m your host, Fran Sorrentino, Client Partner in Financial Services at PublicisSapient.
This episode, I’m joined by Aubrey HB Director of Advanced Analytics for Nationwide Building Society and my Publicis Sapient colleagues, Zack Scott and Simon James. Hello everybody. Today, we’re going to explore the uses and effects of AI and ML in banks today, where we are in the journey of adoption, and what’s next. So Aubrey HB, thank you so much for joining us. Where are you calling in from today?
I’m calling from my home in Newbury at this exact moment halfway between Swindon and London, which is a convenient place to be given Nationwide’s dual offices that we’re currently operating.
Lovely. How long have you been at Nationwide Aubrey?
Since April of 2019. So just about 18 months so far, so good, so far. Great. I love the UK.
We’re glad to have you. I have a lame introductory question for everybody which is what is your favorite real or fictional AI powered personality
For me, I would have to say ex Machina if you have ever seen the movie, it was a very interesting for Ray into really understanding the term test and being able to really identify when a computer takes over versus is actually human. So I thought it was a really, really good science fiction movie that addresses the more challenging themes of understanding AI versus human interactions.
That is so much of a more sophisticated answer than Rosie from the Jetsons, which was mine. So Simon James. Welcome. tell us a little bit about where you’re calling from and I’m going to need your favorite AI personality.
I am calling from today, leafy Surrey. And my favorite AI, I’m going to change the question slightly by answering it metaphorically. And I think the, the best AI powered personality is Raymond Babbitt from the film Rainman, because I think we all saw his amazing computer, like processing power in his brain for being a savant, but also his childlike naivety, which I think is a good metaphor for AI today.
Great answer, Zack Scott, thanks for joining us. Where are you calling from and who is your favorite AI personality?
So I am calling from my kitchen, which is in in West Hampstead in London. And I guess I just watched a movie, which wasn’t great, but it was called Jexi. And what I liked about it was that the AI was not necessarily, you know, understandable by humans. But in the end did kind of help create a better outcome for the main character so that not all kind of algorithms actually are understandable, even if the outcomes might end up working out in the end.
Great answer. All right. Well, let’s dive in for our listeners. You know, I think the first thing we want to do while we’re together is just talk about what the state of play is. And before we jump into that, I thought it’d be useful to just set the table. You know, there’s no possible way you can live and listen to the news and not hear AI and machine learning all the time. Algorithm has gotten a lot of bad press lately, but Simon, can you kick us off with some definitions of those terms and in simple, in simple terms, what it means for financial services?
Yeah. I mean, you know, everyone has their own definitions, but broadly machine learning is really a subset of artificial intelligence and machine learning is the way that machines can, can kind of learn from data without explicitly being told what to look for, what to do, and they can make predictions tirelessly fast accurately. But, and I can do that across numbers and, and language and, and, and video and images in a way that computing power lets us do today. That was never possible like 10 years ago. Artificial intelligence also includes a kind of an element of decision-making and that’s the contentious part really because January we were safe when computers make predictions, but humans ultimately make the decisions when you start delegating their decisions to, to a computer, like an algorithm in the, kind of the recent GCSE and A-level tests.
Then that’s where you know, the edge cases, people who are miscalculated with type one and type two errors, false positives and false negatives, you know, that, and what’s, what’s the cost of that. What’s the human cost of that. And, you know, machines, aren’t very good at calculating those types of things. So that’s the real difference. ML is a subset of AI where we’re starting to see it in financial services will obviously anywhere there’s a process that was carried out by humans that took a long time. That was, that was prone to error. You know, that’s what machine is pretty good. I can do it faster, more accurately. As long as it’s kept to a simple task, you know, so that’s generally where the first place where you see kind of intervention and then over time, more complex things, either technically complex, like, you know document checking online where, you know, we hold our driving license up to the, to the camera, you know, checking that is you. And it looks like the, on the camera, you know, that’s where AI is using computer vision to do those things. So definitely seeing a lot of attraction from those kinds of areas.
Yeah. That makes a ton of sense. And across, you know, the financial services organization we work with at Sapient, you know, what is the most successful application? Is it processes, you know, what, what are the conditions to make AI successful in financial services?
Well, I think today, you know, where there’s a known cost and you can make an improvement in that cost, then everyone’s happy about that because, you know, there was a big cost and now it’s a smaller cost. And so that can be tracked. I think some more of the futuristic edge cases like the value is unknown yet. It’s still in a kind of very experimental kind of mode where you know, the opportunities might be much bigger than the cost saving, but there’s less faith in, in producing the returns on that. And therefore progress has been a bit slower. Obviously everybody loves it where there’s, there’s a black and white outcome to something. And therefore we, we tend to over-index on things that we can easily measure in life. And this is one, you know, good example of that.
It’s a ton of sense. Aubrey, tell us a little bit about where you’re at, where you’re making headway in Nationwide.
Well, we just recently started to build out our advanced analytics capabilities when I joined. So we’ve kind of dabbled a little bit in training up our data scientists in areas of neural networking. And so one of the original use cases that we did was actually identify footfall traffic of some of our branches using CCTV camera and, and it’s really for us at this stage, it was just experimental to identify that we could do it. That on the flip side though, we’ve been doing a lot of machine learning in the space of process optimization and cost reduction and understanding data flows between different pieces of hardware, infrastructure systems in order to really optimize our processes and reduce the costs and the error rates that could happen when you’ve sort of done a build up of a legacy estate, which is very, very common in the financial industry, as well as pretty much any industry that was born or a company that was born prior to the internet being born. So for that’s really where we focused, we’ve also been looking at applying machine learning techniques into a lot of our fraud detection cases in order to improve our ability to track when fraudulent activity is happening in the financial crime space. So we’ve been really focusing there first and foremost,
Simon, it’s really helpful for me when you talk about, you know, what’s, what’s hype and what’s just good data and analytics practice. Can you talk a little bit about that? What, what are the game changers that are accelerating AI progress if there are any, or is it just about good machine learning?
Well, it’s definitely think good machine learning is at the core. You know, if statistics is at the core of machine learning machine learning is at the core of AI, you know, as a practitioner, it’s always really important. You’ve got to track things back to first principles because unless you can work out and if you can logically explain what you’re trying to do with data then, or the alternative is you just have a black box algorithm. You can’t explain it. It’s going to fall over one day in a way that you don’t understand. So as you know, a responsible practitioner of building machine learning or AI, you really got to understand how your model works and be able to explain it to someone, even though the model itself, doesn’t come with all those safety handles and things, you have to go the extra mile to be able to explain it because one of the big challenges is when you’re exposed to data you’ve never seen before.
So, you know, 2020 is a great example. We’ve never lived through a time like this before. Therefore a model has never seen this data and therefore it doesn’t know very well how to deal with it. I’ll be very accurate. So, you know, that’s one of the changes in terms of, so fundamentals never going to go away and everything’s gonna be built from the bottom up, never once the moonshots, but actually the fundamentals, you know, are the things that keep everything going. When people like generally in any feature film, there’s a sense of general intelligence, a robot that knows everything. And that’s probably the bit that’s furthest away. You know, there isn’t a general intelligence that can solve every problem. It’s generally when we start to break down and solve problems on a one-to-one level, individual problems can be solved. Like I could maybe help you, you know cancel your debit card and find a replacement if you lose it in a chat bot, because it’s a very narrow set of questions that you’re going to have.
I can predict what those questions are going to be. And actually behind the scenes, there’s already a manual process or an automatic process that does that. And all we’re doing is putting up a front end, a customer friendly user interface that allows you to get to a positive outcome faster. Now, those things are all eminently possible and can be dealt with today. It’s more like fight. And if you can be 10 grand, I’ve got robot, that’s going to invest that for you. And in a year you’re going to be a millionaire. That’s probably not true. And it’s probably not going to happen because of the complexity of that. And, you know, the kind of the guarantee of certainty that just isn’t isn’t possible.
Well, that’s very disappointing and his podcast is over. Aubrey, tell me a little bit, you know, does that jive with you, as you know, both with your eyes on Nationwide and what you see across financial services, is that true for you as well? Is it that point solution focus?
Absolutely. I think one of the things that we’ve been really concentrating on with Nationwide is actually getting the right minds behind the people in coding these algorithms. I’m very purist in ensuring that all of the data scientists that work at Nationwide are really top-notch technical experts in mathematics and statistics, because it all goes back to that transparency issue. It’s all the audit. That’s actually, one of the reasons I came over to work for the financial services industry in the UK is I really liked how regulated it was and requiring the fact that all deceased Cision’s needed to have a potential human intervention. And it’s only when you can explain the algorithm at the end of the day, that somebody can determine whether or not the decision was accurate. So for us, we’re entirely focused on ensuring that the mathematical processes that sit behind our machine learning and AI are sound, are well articulated and are transparent for any and all use cases. And that becomes a lot easier when you’re really focusing on a single solution. Instead of, as Simon said, a black box version that would, hopefully you would, you would hope would actually solve multiple and many varied kinds of use cases. So for us, it’s really start small grow out and ensure that what we have done is accurate and able for us to explain it to the regulators as well as to our members.
Yeah, I, that makes it sound a sense. And I think those checks and balances as you’ve described are critical with AI and what I’m continuing to hear more of is the ownership, the responsibility of, you know, the ethics of this tool being with the creators. And not necessarily with somebody who’s going to check somebody else’s homework.
All right, well, let’s move on from sort of where we are today to what might prevent us from progressing and maybe what we can do about it. Simon, you know, there, we’re talking about regulation, we’re talking about, you know, the guard rails that need to be really explicit for financial services and the security and safety of, of members and colleagues. Can you talk a little bit about the comfort level you see across enterprises? And do you think some of the, you know, recent press about badly applied AI affects that comfort level?
Yeah, well first start with, you know, legislation is the floor, not the ceiling, you know, we shouldn’t aspire to be in line with legislation. That’s the minimum element we’ve got to aspire higher and the trouble is, is for many walks of life and technology law at the moment. The law is struggling to keep up with the pace of change in digital. So, you know, obviously the last couple of years we’ve had the GDPR legislation come in, you know, it’s been 20 years since that Lord properly been updated prior to that. And now the world has changed in that time. So, you know, there’s a constant arms race with trying to keep ahead of legislation, but you shouldn’t aim to be, you know, legally valid. You need to go much higher than that. And the trouble is there’s a big difference in what we can do and what we should do, because we can pretty much do anything these days with the computing power and the data that we have, but there’s as much SIM slim, a set subset of things that we should be doing with the data that adds value for the customer, does it in an ethical way.
And you know, it doesn’t lead to bad practice. And one of the big challenges is unintended consequences of what we’re doing, you know, so we could, we could build a credit
Risk model and improve that, but we might be tuning out a subset of the, of the, of the country that you know, would con no longer want to give credit to, you know, and that might be biased, you know, and therefore ensuring that we’re not being biased and checking for those things is, is, is important. But the things that holding it back is really things like reputational risk and the risk of things going wrong. And that’s why we, you know, when you begin, you must really try and focus on things, you know, minimizing, mitigating that risk because otherwise it’s one step forward and two steps back.
Yeah. It makes a ton of sense. And I know you and the team have done a ton of work on [inaudible] biased AI and the ingredients of doing that really fairly, which might be a whole other podcast. Yeah,
Definitely. But I think the first step in that is acknowledged that bias exists and exists from the humans exist in a historical way. We’ve treated that people aren’t, you know, what we’ve done with them. So like, and, and as you can recognize the bias, you can’t go to step two and you can’t begin to start solving for
Yeah. Agreed. And certainly the, you know, the conversations we’re seeing in the media about those biases might, you know, create a little bit of a fear factor for progress, but, you know, Zack, I’m interested in your perspective on this, is it sort of the hype or is it, you know, the way that organizations are set up to deliver AI and ML that you think really hinders progress?
Well, it’s a very good question. I think a lot of focus when it comes to data and I, AI com tends to come on, you know, the quality of the data and getting into one place. So then, you know, the actual technology and kind of other capabilities required to use that data effectively. What I tend to see actually though, is that the kind of major challenge is actually beyond that. And it’s in converting the insights that you generate from your AI and ML, and actually like delivering that and bringing that to life for your clients, for your colleagues. And I think the realization that a lot of firms are having is that unless you’re better connected as an organization, unless you are better able to connect, you know a new and improved model to the different workflows that need to change to the way that people need to change their roles to the way that they need to change how they interact with each other and with customers it, it, doesn’t kind of matter how, how smart you are if you can’t bring it to life. And so a lot of the changes that that I’ve been seeing is, is that realization and the realization of the need to, as you say, eliminate some of those organizational silos to be able to go, not just from a great idea, but actually bringing it to life and bring it into production.
Yeah. And is it an education question? Is it an access question? How would you, you know,
I think it’s, it’s, it’s kind of all of the above. I think there’s a bit of it of kind of better connecting the different folks so that you don’t have a data and AI team over here doing some great stuff, but then a kind of customer facing, or colleague, you know, a backend operational team over there doing something else and them not being connected. I think the educational part comes with the fact that it’s always going to lead to some form of change and change always will meet with some form of resistance and always requires some degree of education. And in particular, you know, when you’re talking about things like AIML, you might be kind of overturning, you know, decades of, you know, institutional wisdom that you basically are proving is not correct anymore. And that will meet with a, quite a lot of resistance.
And as part of that, one of the kind of big things I tend to see is that as part of that resistance, there’s an unfair standard setting that gets applied to AI and ML and kind of data-driven approaches in that. It’s almost expected that they’re perfect from the start, or if they’re have any flaw it’s kind of like picked at when in reality, actually, if you kind of turned it on its head and said, well, how well do we actually understand our existing processes and how well do they work? Can we actually say that they are better than this new way that we’re proposing? And often the answer is you can’t and they’re almost kind of protected by that lack of visibility, but I think the realization that you need to overturn that and be open to progress rather than just perfection from the new approaches that you’re taking is quite important.
Yeah. Great point, Aubrey you’re nodding, when Zack is talking about some of those barriers, you know, is tell us a little bit about how your building capability and you know, how you’re introducing this very new kind of talent to Nationwide. It’d be interesting to get your take.
So I think one of the main reasons why I was nodding is there, there are two sides to this coin. The first is not only is information and perspective siloed for a lot of organizations. Data is actually incredibly siloed and that, that makes it quite challenging to do these more profound machine learning techniques. So one of the things that we are currently in the process of doing is establishing the data framework in a new environment and new data Lake that is actually fit for advanced analytical techniques to be deployed on top of the full repository of our member data. And that comes with a lot of thoughtfulness in terms of how we want to bring that data together, starting with the logical data model and ensuring that all of our metadata is encoded in machine readable and human readable sorts of ways. Now, along side, that in parallel, one of the things that we’ve been really doing in terms of the advanced analytics team is really working very closely with every single of our business stakeholders and helping explain every single pathway and every point of decision-making and helping possibly moving a little bit slower at times, but helping them understand the techniques that we’re actually deploying and have some comfort into these new approaches.
And that is actually been one of the bigger challenges is really getting people who might not have been traditionally data literate to join us on that data literacy program in conjunction with the work that we’re doing for them to help them understand how these methods can solve their problems in a, in a better way. But I, I do think, and I was going to add in that some of the challenges is also around an organization having the expectation that this can be done really quickly, you know, Oh, you’re using machine learning. So you’re going to be just as readily available to give us answers as the likes of Google or some of these places that have been deploying these techniques for many more years than we have been. It’s, it’s really helping people understand that these are really hard types of algorithms and it does require complicated mathematics and getting the actual backbone of the math and statistics right. Is worth the effort and the time delay that it might take to solving some of the problems at this stage.
Yeah. I think the you know, there’s a whole movement of data democracy going on. It’s not the job of a bunch of guys with lab coats sitting in the corner, solving all the world’s problems for us. It’s part of everyone’s job going forward and completely agree with Aubrey around. We want to raise the bar in terms of everyone’s understanding an IQ around data, and it’s nothing to be fearful of. It’s something that’s, you know, in all of our lives in many different ways, you know, these days, we all look at how many likes and follows and, you know, we’re all analyzing our own social media presences. So we kind of use of data in our lives now. So you know, it’s not the job of data people. And I think Zack said earlier, you know, oftentimes no data person, just a role in a cross-functional team solving a business problem for the benefit of the company and our consumers. So, you know, it’s not a S it’s not another new silo is something that will benefit greatly from being you know, brought into the general populace as much as
This is a question for everybody. If you could change or put one thing into place, unlock more effective, more widespread use of AI in financial services, organizations, or Aubrey and Nationwide, what would it be? Zack I’ll start with you.
Mine, like a kind of small one, but one thing which I get really excited about is when folks can make AI and machine learning accessible to people who aren’t necessarily as deeply going a data literate. So like you don’t need like Python coding to kind of play around it and understand the implications of it. So as an example, some of our colleagues Publicis Sapient have this you know approach called not knowledge as a service and the way they’ve set it up is you can like go to a website, put the URL into it, and it will automatically digest it and convert it into a FAC or a searchable, you know manual that you can use. And what I like about it is not necessarily that it’s like really complex AI or ML, but what it allows you to do is to showcase that to someone who isn’t necessarily as you know, data knowledgeable, and they can engage with it, understand it, start to think about, you know, what are the applications that I can see using this for. And I think having little elements like that, that can bring it to life of can probably go a long way to helping adoption
Love that. And did you say FAQ instead of FAQ?
I did, is that I, I’m not really sure what’s the proper, this is an American thing or a British thing, or just a Zack thing that it’s called a fact for a second thing.
What I’m adopting it from here on out. Aubrey, how about you? What’s one thing you would put into place.
This is going to sound completely left field, but actually better teaching of math and the applications that math can be applied to at lower level educational level. So in, in grade school, we call it middle school and high school, but getting people to understand how important it is to study the more STEM fields and have that appreciation early days. And then that way that ideally will translate into folks that have that more data literacy once they land into a career without having to go through the efforts of trying to retroactively convince people that math is a subject to study at some point in their lives.
Love that answer, Simon.
Yeah, I think rather boringly probably results. I think like unambiguous results, success has many authors. So as soon as people start seeing results coming in, you’ll find a lot of motivation to get behind some of the engagements. And that’s maybe the frustration today is that still people are struggling to produce not, not kind of results from a prototype or a PO a proof of concept for, I say POC. But you know, kind of scalable results that work over a long period of time. That’s what people need to buy into, I think. And so, you know, the more we can do to, to share and, and show the results from the work that we’re doing that will accelerate adoption. I’m sure more than anything.
Yeah. That feels like a virtuous cycle with the other answers as well. You know, the more results, the more we can talk about, you know, what we’ve learned, making people literate, making it democratized, that makes a ton of sense to me as well. So now let’s pivot to what’s ahead. I’m interested in, you know, people are putting their energy in the immediate future and maybe what they’re hopeful for in the future. Aubrey, what’s your focus for the next three to six months
Next three to six months, it’s really establishing our foundations for our data fabric to ensure that we actually have an environment that we can deploy these more complicated methodologies, but actually in terms of deploying real solutions, we’re focusing on bringing robotics into data quality checking in terms of our customer facing member facing systems, just to make sure that the data that we do have on our members is as accurate as is possible, so that when we ultimately want to use it, we can use it without fear or concern that it’s going to have an adverse effect on our models and our machine learning techniques.
That’s great. Great to hear Zack, you know, when you’re thinking about advising clients do you, you know, do you encourage them to do more of what they are doing? Or do you think they should be testing more going broader with more methods?
Probably all the above, but I guess one of the things I guess hitting, picking up what Simon had said of like needing to show results, I think is quite key. And, and I think the big thing that I’ve seen is there’s a lot of like big ideas about data analytics and what it can do. And people have spent like a lot of money on lots of different elements of kind of bringing that to life. But I think what it really does take sometimes is a pretty ruthless focus on bringing to life a couple of really great examples. And by that, what I mean is not, you created a proof of concept and not like, okay, we had generated this unique insight over here, but actually bringing it all the way to life, you know, to live to clients, to in the organization and having an actually up and running. I think getting even one little bit, you know, one little example all the way through can actually reveal a lot about how you need to change that end to end pathway so that you can get more and more of that through. And I think oftentimes clients get kinda overwhelmed by lots of different possibilities rather than just being a bit more focused on a few really important ones and making sure they get delivered.
Yeah. I think that brings home nicely the importance of creating a business. Who’s literate enough to understand the value of breaking down the silos, getting the data into place, proving it out end to end. And so matter, you know, it doesn’t quite matter how big it is so long as it is end to end in credit.
And as part of that is like a lot of that is actually going to end up being kind of beyond just the data and AI, right. It’s going to be linkages with the technology teams, it’s going to be changing the kind of the functional usage and the roles and responsibilities. And so it’s the realization that actually to do that, change the AI and ML and kind of data elements of it are a component, a key component, but only a component and, and kind of bigger hole that needs to be delivered.
So you’re saying we all have to hold hands and do it together.
Gotta do it together.
Excellent. so we’re coming to a close I thought I’d get a few last thoughts from you all. And I’m so grateful for your perspectives if in the future budget were not an issue and you have the team of your dreams, you were ready to start tomorrow in Nationwide or any financial services organization. What is, you know, the end to end project you would work on Simon?
I think topically post COVID, I think, you know, how we can help entrepreneurs get kick-started and by going and like, it’d be great for, to use AI to kind of review their business cases and give them the right insight to help them flourish. So not only just come from money from, or, you know, funding from a bank, but they come from business support, I had to, to enable them and be partners in true growth. And that’s important for the economy and important for society post COVID
Zack,
I’m going to be boring and copy Simon a bit. I’m very much caught up right now in, in kind of lending and data and AI and, and on how that can play a role and kind of all elements of the lending process. And specifically, I think where I see a lot of major gap is for small and medium enterprises and even large corporates of how can financial service firms better serve them. And then also, you know, make better decisions about how they lend to them and how they manage the, the money that they have went already.
Aubrey, how about for your dream project?
It’s really bringing a human touch to what could feel like digital interactions and really helping us find the best way to service every type of our member, because we have vast and varied members with different issues and concerns and financial considerations and just building a holistic environment that allows us to service everybody equally to the best of our ability, even when we cannot, you know, stick a financial planner for every single one of our 15 million members, it’s really bringing some of that into computing in a way that it offsets the human component of having to do that service. But with that same Nationwide caring, support, understanding, and making sure that every interaction that we do have with our members is the best it possibly can be.
All right, a couple of rapid fire questions, and then I will bring us home. I’m going to go all three for each of you, and I’m gonna start with Zack rapid fire. One thing you’d change to data and AI easier, one word to describe what it takes to succeed. And then a moment in your day to day work that gives you energy.
So one thing to change to make data and AI easier would be better data readily available.
Francesca Sorrentino (33:02):
One word to describe what it takes to succeed using data and AI in financial services
Openness,
A moment in your day to day work that gives you energy.
I really liked the morning. I’m a morning person. And so I have a nice like morning tea. And then I usually will start my day with a couple of catch-ups with, you know, the people I’m working with most closely. And that, that usually energizes me
Aubrey, over to you. One thing you'd change to make data and AI, easier
Metadata wrapped around everything. Yeah.
One word to describe what it tastes it takes to succeed in data and AI in FS
Persistence,
And a moment in your day to day work that gives you energy
Hearing gratitude from our internal clients,
Simon.
Um I think making data and AI more explainable is core. Things are opening up to more people. I think the key thing that you need is patience. Doesn't come immediately and the first go round isn't necessarily the best. So I think patience is a virtue that you definitely need. And the thing that triggers me is I need to find a riddle to solve, or I'm very curious. So once I find something that I don't really understand immediately, that triggers my, my motivation to go find out more about something. So I need it. I need an enigma or a riddle
That is so on brand Simon, James. Everyone, thank you so much for your thoughts. I think this is a really valuable, you know, conversation around where we're at really what's coming down the pike and not just what it takes to succeed in using this really exciting toolkit, but lessons for life openness persistence, and give Simon Jim riddles. Thank you everybody keep an eye out for more XBank podcast episodes in the coming months, we'll be discussing making the most of cloud and how to build a better neobank subscribed to our dedicated XBank website at xbank.publicissapient.com, where you can also learn more about digital banking transformation.
Francesca Sorrentino
Client Partner, Financial Services
Zachary Scott
Associate Managing Director, Management Consulting