AI in Banking: From Hype to Measurable Outcomes

Jim Marous (00:11):

Most banks, including potentially yours, are falling into a dangerous trap. They're announcing AI initiatives, holding innovation showcases, and talking about digital transformation. But when it comes to measurable business results, 70% have nothing to show for their investment. That doesn't mean that the investment isn't paying off, it may just mean the measurement's not taking place.

Jim Marous (00:36):

The winners like JPMorgan Chase, Capital One and a few others aren't just lucky, they've decoded a specific approach to talent, execution, and measurement that lets in turn AI investments into strategic advantages. Meanwhile, many are still wasting budgets with little or nothing to show for it beyond PowerPoint slides.

Jim Marous (00:58):

Today, we're joined again by Alexandra Mousavizadeh, CEO of Evident, who has just released the most comprehensive analysis AI outcomes in banking. Her team tracked 173 use cases of AI across 50 major banks, and the data reveals a stark divide emerging in our industry.

Jim Marous (01:20):

Alexandra will reveal which banks are already using AI to steal retail customers, how they're measuring real ROI, and most importantly, what retail banking leaders need to do differently in the next 12 months to avoid being left behind.

Jim Marous (01:36):

If you're responsible for retail banking strategy, customer acquisition, digital transformation, or just the customer journey, the next 30 to 40 minutes will determine whether your bank thrives or merely survives in the AI revolution.

Jim Marous (01:50):

So, welcome back to the show, Alexandra. Before we begin, could you reintroduce yourself to those who may not be sure who you are or whoever it is?

Alexandra Mousavizadeh (02:00):

Absolutely. Thank you so much, Jim, for having me on again. You had me on the first time just when we had launched the ranking of the banks on the AI adoption, and here we are two years later. But very happy to just reintroduce myself to those who didn't listen to the first podcast.

Alexandra Mousavizadeh (02:19):

My name is Alexandra Mousavizadeh. As Jim said, I am the co-founder and CEO of Evident. And Evident is a benchmark business where we rank financial services on their AI adoption progress. And we rank the 50 biggest banks in North America and Europe, which we've done for now three years. And we just recently released a ranking of the 30 largest insurers in North America and Europe just two months ago. And we will be releasing a payment platform index and an asset management index and a wealth management index, and so on.

Alexandra Mousavizadeh (02:57):

And we might venture into sectors outside of financial services at some point. But the financial services are just really a rich area and sector to measure because there's so much investment going into AI. There's so much data, there's so many use cases, and while it is a regulated sector, there still is a lot to do and a lot that the banks can do, and there's a lot of competition.

Alexandra Mousavizadeh (03:24):

So, it is a bit of a, probably a winner take all. And to your point before, Jim, what we're seeing for sure is that the gap between the leading and the lagging banks is only growing ever bigger. And so, you definitely see that if you're leading a leading bank in AI, you are ... what we are seeing in the data at least is that they're only doubling down and they're pulling out much, much further ahead. And so, therefore, that gap between leading and lagging banks is only increasing.

Alexandra Mousavizadeh (03:57):

And my background, I've been in the index measurement business all my career at Moody's, at Morgan Stanley. I've run a rating agency, but have for seven years now measured AI, adoption and deployment with the release of the global AI index back in 2018 (can you believe it) where we measured nations on their AI progress which is still in existence. But when we got tapped constantly from the financial services to ask if we could adapt our methodology to financial services, we pivoted and set up Evident. And yeah, that's almost three years ago.

Jim Marous (04:38):

So, your latest report on outcomes shows that 47 of the 50 largest banks announced 173 new AI use cases in the past year, but the real startling result, which kept on intertwining your whole report, was that only 30% report actual outcomes. So, what's fueling the disconnect between AI announcements and tangible results is simply a reporting issue or is it maybe that they're not even measuring the results yet?

Alexandra Mousavizadeh (05:09):

So, maybe just to step back, I mean the outcomes tracker that we've built is based on what the banks are making visible. And so, you'll have some banks that are comfortable reporting on everything from what's the actual AI spend to what is the number of use cases to what is the actual ROI in dollar terms. Very few banks do that. JPMorgan does that, DBS does that, and a few others.

Alexandra Mousavizadeh (05:37):

Then you've got a cohort of banks that are reporting use cases with an associated outcome, which is the ones that you were referencing. And then there were banks that might take a slightly more cautious approach or might not be that far advanced in sort of a central orchestration of measuring impact and sort of the whole gamut of ROI.

Alexandra Mousavizadeh (06:01):

So, I'm more than happy to talk about the use cases and what they're doing in the bank because there's so much pressure now from investors, shareholders, and also talent to see what is the bank actually producing. But not maybe quite there yet in terms of articulating an impact, in terms of a time saved or a dollar value or revenue uplift, because once that cat is out of the bag, that number will be tested over and over again.

Alexandra Mousavizadeh (06:33):

So, it's not that the ROI is not necessarily there, and we can see it from our private benchmark that we do with the banks where we get all of the data from the banks on their actual ROI and all of their use cases.

Alexandra Mousavizadeh (06:47):

So, the split is what are you making public? And their banks are slightly more cautious versus sort of what's actually going on behind the scenes. And there you can see there is actually impact, but you just maybe want to have it run for a little bit further or you're seeing efficiency gains that you want to actually translate into a dollar value, but you might want to have it run for another six months.

Alexandra Mousavizadeh (07:14):

So, we're still very early days, we shouldn't forget this. We're still very early days and shifting AI use cases from ideation and into production. And then from there to measure the ROI is only just really starting to have a consistent methodology to measure it in a way that's constantly monitoring the progress, and so on. And there's a long way to go. We're just in the foothills of rolling out AI, so we shouldn't forget that there's a very long way to go.

Jim Marous (07:45):

Well, it's interesting, we were talking before we got on air that MIT just came out with a study that looked at the outcomes also, and they came up with very similar results as far as the trending, that there's a lot of initiatives that have not (and I got to be very careful about this because I think it can be mis-defined) recorded a return on investment yet. And I hate to have financial institutions that are tentative about AI, thinking that means they're not generating results or they're not generating positive results.

Jim Marous (08:17):

But again, I think it's a situation that we have to be very careful about the click bait versus the reality that many are still in the process of understanding how best to measure that. What was your takeaway from the other study?

Alexandra Mousavizadeh (08:31):

I think two things about the MIT report. I mean, directionally was probably accurate. There is actual, tangible financial returns are not fully coming through yet, A because there's a lot of testing still going on. There's a long way to go in terms of getting use cases into production across functions and lines of business, and were just in the early days. And as I said before, early days on measuring the revenue uplift or actual cost savings.

Alexandra Mousavizadeh (09:08):

So, the MIT report was probably directionally right, but it was also very flawed. They only looked at the financial gains, they didn't look at other things like risk reduction and customer satisfaction, and hours saved that were then used for other things that weren't necessarily translated into dollar values, but would probably be things that would come through in growth of market share over time, but not taking that into account.

Alexandra Mousavizadeh (09:38):

And the other issue was that they looked at use cases in production and ROI only in a six-month time period, which is not enough time to actually identify a realized return. So, either, it was a lack of understanding of when do you actually expect to see true value or it was just a sort of a fault in the study, and therefore, I think we should ... it's probably more than 5% that are creating value, but-

Jim Marous (10:10):

It's just how you measuring it. Yep.

Alexandra Mousavizadeh (10:12):

Yeah, but it's not 50%. And the interesting thing was also that it had such an impact on the markets, and I know it has been debated a lot when it came out a couple of weeks ago, but that is because the markets are skittish in terms of looking for the evidence that the adoption and the AI deployment at enterprise level is actually happening.

Alexandra Mousavizadeh (10:36):

Because otherwise your Microsofts and your NVIDIAs and so on that are delivering the tools will not have something to base the high valuations on if adoption is in fact going to take a lot, lot longer, and the ROI going to take a lot, lot longer than people had anticipated.

Alexandra Mousavizadeh (10:55):

So, when a report like that comes out, it is going to spook people a bit.

Jim Marous (11:00):

So, your report showed that the outcomes are finally moving beyond efficiency and back-office gains to include revenue growth, which you found had increased to 16% of use cases in the first half to 2025. What's driving this change from what was, I'm going to call legacy AI utilization and fraud risk and back-office outcomes and efficiency to the types of use cases actually contributing to a better customer experience and revenue growth.

Alexandra Mousavizadeh (11:30):

Yeah, I would say it's still a relatively slow shift. I mean, it's a gradual movement away from solely seeing returns from back-office, low risk internally facing gen AI use cases, which is in the sort of the pipelines of the bank and the functions and the IT functions and fraud and KYC and so on.

Alexandra Mousavizadeh (11:55):

Well, that's not fully only back-office of course, but that's sort of where the deployment was a natural place to be initiated because it wasn't customer-facing with the lack of reliability of the large language models, whichever one you used, you couldn't let it be fully customer facing. You still can't do that.

Alexandra Mousavizadeh (12:16):

So, deploying it into the back-office functions was a very natural way and place to start. You want to try and see, can we get our cost to delivery down? Can we make our operations more smooth? Can we take out drudge work that the analysts are doing? Can we pull in all of this data so we can deliver better analysis and put tools in the hands of our investment bankers, our wealth management analysts, our marketing people, and so on. But there will always be a human in the loop.

Alexandra Mousavizadeh (12:51):

And so, that hasn't written out. I mean, if you ask any bank really about their KYC process, the idea is that you get to sort of an 80% of the KYC process automated, but there's no bank that has reached an 80% automation of KYC of knowing your customer and onboarding your customer. They'll get there eventually.

Alexandra Mousavizadeh (13:13):

So, this is still a lot of room for gen AI use cases being deployed in functions across the back-office still, and there's a lot to be done on things that are more customer-facing. Now, the retail banks have a lot of AI already in their customer engagement, but it is very carefully deployed, a lot of guardrails, you've got human in the loop, but you do see that functions are being tweaked and tuned. They're being improved on the margins with gen AI tools.

Alexandra Mousavizadeh (13:46):

And that is what we're starting to see, sort of there's more and more comfort with that, and that is moving. More and more AI tools are deployed for customer engagement purposes, and so on. And sort of, you're starting to see sort of in areas like trading where you're seeing deployment of gen AI that are catching mistakes, fixing them quickly, the pickup and pace and speed of trading and refining on the edges that are leading to sort of proper revenue uplift as well. But there are many examples.

Jim Marous (14:18):

So, when you look at the internal versus external, you found that 85% of gen AI use cases are still limited to internal users, you've mentioned. What prevents banks from your perspective, from deploying generative AI and AI solutions more externally? And when do you think this will change?

Alexandra Mousavizadeh (14:37):

Well, it comes down to governance at the end of the day. And it comes down to the risks — and I'm not talking about AI risks, but the risks that banks will be willing to take with their customers of trying out sort of the deployment of more sort of AI-driven tools.

Alexandra Mousavizadeh (15:01):

Will they come up with an odd answer? Will it be slightly kooky? Will it be a bit weird? But will it be a bit off? Are they suggesting things that are a bit askew?

Alexandra Mousavizadeh (15:09):

In addition to, of course, the risks that are there with making decisions if you just leave it without a human in the loop to make decisions on loans and credit lines and mortgages and so on, which I think we're a really long way for having an automated application and decision made fully on sort of bigger ticket items without a human in the loop, for sure. Not the smaller ones are already very ... so, you can sort of bucket it into sort of two buckets, which is like the deterministic and the non-deterministic AI.

Alexandra Mousavizadeh (15:43):

The deterministic AI is tried and tested. We know that's still going to move forward, if this, then that. But things that are non-deterministic, that can still produce slightly wild answers, we will need to probably see it through a period where they get properly fitted in, so there's no loose answers. That they're properly fitted into the data, so it's just fully only responding to the data that it's supposed to be responding to.

Alexandra Mousavizadeh (16:14):

And then there's the governance aspect. This is one of the biggest hurdles when we ask the banks around their time to production. From an ideation to getting a use case into production has come down drastically when you're looking at low risk internally facing use cases. But when it comes to higher risk externally facing use cases, that time to production is still high.

Alexandra Mousavizadeh (16:38):

And the reason for it being higher are threefold, one is data infrastructure, legacy banks, data is still everywhere. There's a lot of that process and platforming and organization that needs to be done.

Alexandra Mousavizadeh (16:51):

The second issue is always around resistance. Resistance to changing workflows. Don't want to really do that. If I'm a head of a line of business and I think things are functioning great as they are, do I really want to rethink my whole process flow and put gen AI or agentic AI workflows through it? I may, but I may not want to do it today because I'm busy. I might want to do it next month or next year.

Alexandra Mousavizadeh (17:17):

But the third issue, the third hurdle is always around governance and especially the model validation side for gen AI use cases, that is a big hurdle to overcome. And banks are still very much wrestling with what that might look like. And that goes hand in hand with talking to regulators about what would they be comfortable with, with model validation of a gen AI use case. You validate it and by definition never come up with the same result, which is what you need to have in your model validation that it is tested and comes up with the same results. So, among many other things.

Alexandra Mousavizadeh (17:53):

But that second line of defense of the model validation is one of the biggest hurdles to overcome when it comes to externally facing gen AI or agentic AI use cases.

Jim Marous (18:03):

Well, it's interesting because it's the fear of making a mistake. Banking's always been on a, I'm either alright or I'm all wrong. And yet every way a customer's coming in contact with AI and gen AI and agentic AI outside the banking industry realizes there's going to be slight errors.

Jim Marous (18:21):

Heck, I'm going down the road last night and we're reading our GPS system, and it wasn't perfect yesterday, so it didn't know about a couple traffic jam, things that you get very used to. But if you are looking for the perfect solution, which banking often tries to, you can be stuck in looking for perfection rather than something that can move the needle forward.

Jim Marous (18:43):

And I think, especially as we start talking about agentic AI, which is kind of like doubling down on what you think your answer's going to be from an AI standpoint — your data shows that only 9 of the top 50 banks have a documented agentic AI use case with just three: Capital One, JPMorgan Chase, and Bank of New York detailing supporting architecture.

Jim Marous (19:07):

What separates the mindsets or the organizations or the way they do business among those that are starting to dip their toe in that agentic AI use case, which again, doesn't have to be the answer to every transaction, every activity a customer takes, but you can use the tools still to move forward. What sets these organizations apart from your perspective?

Alexandra Mousavizadeh (19:30):

Yeah, I mean, it's all about having — I mean, the banks that you mentioned, what do they have in common? They're first of all … the way we measure, and they show up very strongly on our index, they're very high performing banks. They have a very strong talent pool. They have a great experimentation mindset. So, let's experiment with this.

Alexandra Mousavizadeh (19:55):

We can see a workflow where agentic AI systems would fit really well on it. Let's try and test that out. And then narrow projects, but they're testing it out enough so they can go out and say, “We've tested this, but actually works.” We have a specific use case where we've seen agentic AI be a really good tool set for this. We've worked out how we get the data in place so it can run over it, and we can apply it. And here's an example of how this would work in our bank.

Alexandra Mousavizadeh (20:39):

And so, it does take maturity to be able to onboard something as new as that at the time (this is already, six months ago, in the world of AI that's a long time ago), and now most banks are experimenting with agentic AI workflows. And so, it takes a maturity and that sort of confidence to know what is your process with which that you test, and can you get to a result pretty quickly enough so you can make it visible?

Alexandra Mousavizadeh (21:12):

And the making it visible is also about putting a stake in the ground. We're so great and we're very good at AI. We've got really strong teams, we want to be showcasing that we're on it in terms of those innovation breakthroughs. So, the next one might be quantum, where we will see quantum inspired solutions to use cases in the banks. But so, you want to be showing that you are on the front foot, and that you can test this and come up with a solution and showcase it.

Alexandra Mousavizadeh (21:47):

Also, for talent attraction purposes, also for demonstrating that AI is top of your mind and that you are deeply embedded in the emerging technology. But what sets them apart is that central orchestration, extremely strong talent stack, innovation mindset. And all of these banks have in common that they pull the innovation levers that are there for them.

Alexandra Mousavizadeh (22:16):

They've got great AI applied research teams, they've got a lot of patents in and around these topics. They partner with a lot of universities. They invest in AI companies. And so, all of these are parts of the playbook that help accelerate the thinking and the deployment of AI.

Jim Marous (22:37):

That's interesting because your report, both this one and the one you published last year around talent really reinforces the fact that as much as the technology is there, it's still going to take AI talent, humans to make it work. Deployment's going to be held up if we don't have a good talent stack in our organization. In fact, last year revealed the direct correlation between the quantity of AI talent and the success of use cases.

Jim Marous (23:07):

Now, we can argue which way which comes first, but in fact, banks with more AI talent report twice as many use cases, and were one and a half times more likely to see an ROI. Is the connection just between talent and outcomes, or does it also include the whole mindset, the culture of generating results from your AI investments, and the fact that more people are involved in it?

Alexandra Mousavizadeh (23:34):

Yeah, Jim, I think you sort of come right to the crux of the whole problem. Well, not the problem, but sort of what is the ... if you were to pick a leading indicator for a bank successfully deploying AI or being AI mature, it would come down to the talent. Being able to attract the very good talent is really a game changer for your ability to move on AI deployment quickly.

Alexandra Mousavizadeh (24:11):

Now, if I just step back, and just for those who are listening who don't know — talent weighs 45% in the index. So, we weigh it heavily. We weigh it heavily because when you look at what it takes for moving as I say, the use cases from ideation to production, it's all about talent.

Alexandra Mousavizadeh (24:32):

10% of it is the technology, but the rest is really the organization, the labeling, the infrastructure of the data, and the talent. Because without the talent, you can't identify the problems, you can't find the solutions, and you can't implement it and put it into production.

Alexandra Mousavizadeh (24:49):

So, we look at 240 role titles within the bank talent stack. And we look at this in the buckets of AI development talent, you've got to have AI model risk talent. You've got to have AI implementation talent, you've got to have AI researchers. And the stronger those buckets are, the stronger that talent stack is, the better your bank will do. Because you need your researchers to go and sit side by side with the business leads to figure out where are the areas that we can improve with AI?

Alexandra Mousavizadeh (25:20):

You need AI development, you need software AI implementation, you need model risk to make sure that you can get it through the governance structures as fast as you can with someone who understands the AI underpinning or the AI model. And then you definitely need the AI product management because otherwise, you can't get your use cases into production. We are also seeing now that the leading banks are hiring physicists and pure mathematicians and so on.

Alexandra Mousavizadeh (25:50):

So, the needs are big and it's very difficult for banks that don't have that deep AI talent stack to move as quickly because of all of that, the identification, the solving of the problems, and moving it into production, the validation. If you've got gaps in your talent stack, in any of that process flow for a use case, it will take longer. And so, over the three years that we've run the banking index, the weight of the talent pillar in the index being 45% has been confirmed over and over and over again.

Alexandra Mousavizadeh (26:37):

So, that's why when you see the banks with the strongest talent stack, and we look at volume and we look at density, i.e. the amount of talent that you have as a proportion of your size of bank, we look at those figures and see that there is that very strong correlation to the outcomes and the ROI. And we see that both in the private data and it very much tracks in our private data as well.

Jim Marous (27:06):

Let's take a short break here and recognize the sponsor of this podcast.

[Music Playing]

Jim Marous (27:16):

It's interesting, the banking world continues to love developing reports, developing insights for internal use, where one department can share something with another department, can find out how they're doing internally. My biggest frustration, especially as my understanding of what can be done with AI, is that I don't feel it as a customer.

Jim Marous (27:37):

And this is an ongoing challenge. And you've indicated in your research, on the talent research, but also in your outcome research, there's a big difference between AI talent that is on the technical data specialist side versus those that are now in the implementation phase.

Jim Marous (27:55):

And my thought is it ain't nothing till it's deployed. It doesn't matter if it's internal deployment or external deployment, but I think you found that the leaders really are focusing a lot more on getting implementation talent. In other words, what do you do with all this great stuff we found out?

Jim Marous (28:14):

And I think that's going to be the key to agentic AI and things of this nature, because the real dollars and cents of this is about how you deploy it internally or externally. And what do you see from that? I mean, what does it tell you about where we are in the maturity wave, not just for the big banks, but in banking overall?

Alexandra Mousavizadeh (28:34):

Yeah, I mean, that's a very good question. Sort of everyone's guessing, where are we on that maturity curve? Are we 5% in? Are we 10% in? It is very clear-

Jim Marous (28:47):

Or does that target continue to change?

Alexandra Mousavizadeh (28:51):

Or does that continue to change? And also, look, we spend a lot of time thinking about what does the bank of the future look like? What does banking look like five to 10 years from now? Are the ones that are leaning most heavily into AI, like your JPMorgans and your Capital Ones and so on, are they going to be fully 100% AI? What does that mean?

Alexandra Mousavizadeh (29:12):

Is every function going to be touched by AI or is it going to be as we say, augmented by AI? Or is it actually going to be many functions that are completely replaced by AI, so you can do a lot more with less. And so, I think that's sort of the big debate that is raging.

Alexandra Mousavizadeh (29:33):

I think what is clear is that also looking at other technical transformation periods, is that it goes slowly, slowly, and then suddenly, you hit a tipping point, and it goes really fast. And we're not near that tipping point yet.

Alexandra Mousavizadeh (29:51):

I mean, if I were to put my head on the block on this gym (and I'm sure you would probably ask me to do exactly that), I would say that we are probably still going to be in the slowly, slowly until end of 27, early 28. And we are going to hit a tipping point by then, and then it's going to be very fast acceleration of AI adoption.

Alexandra Mousavizadeh (30:13):

Because right now, you are seeing use cases going in sort of pretty much everywhere, but it's still quite light touch. In some areas, it's sort of deeper. In many areas, it's just still in its very early days for many reasons. The data is just not strong enough for you to use gen AI tools on it. Externally facing use cases, all of the things that we just touched upon, it means that it just hasn't penetrated that deeply into the banking system yet, but it will. But it is that slowly, slowly, and then very, very fast.

Alexandra Mousavizadeh (30:46):

So, I think that was answering your question about sort of what does the future look like? And it's an incredibly interesting thought experiment which we do all the time. That is sort of, what does it look like five to 10 years from now? And the only way you can really answer that is breaking down all of the functions that a bank does and working out, where are the areas where you can replace and where are the areas that you can augment. And then you can try and come up with some kind of an extrapolation.

Jim Marous (31:20):

And it’s interesting, because more than ever, I believe it's going to be defined by what the customer expects. And I put that in a big umbrella, but the reality is my determination of what my bank should do is determined more by Uber, more by Airbnb, more by Amazon and Netflix and Hulu than it is by another bank, because I don't believe any banks are moving fast enough.

Jim Marous (31:46):

What's interesting is your research looks at the top 50 banks, and there's a vast gap between some and others. I'm not even seeing the biggest and smallest, albeit Chase does have a clear advantage in many ways in the way they do things, and there's certain organizations like the Capital One, you're going to say, “No big surprise there.”

Jim Marous (32:06):

But I think what's more surprising and more eye opening for the industry as a total from top to bottom, not just 50, but beyond that, is that nobody's way far ahead in such a way that, geez, I got all the answers. And it's certainly not divided equally among even the top 50 banks.

Jim Marous (32:25):

I think what's interesting also, what I see, is the innovation, the idea set, the way organizations do it are happening within some of the biggest, and some of the very smallest because they've found a way to leverage third party solution providers to provide them the AI talent on an episode by episode basis even with regard to the management of the data.

Jim Marous (32:51):

These third-party providers say, “Tell you what, we can take your data and make it applicable to an agentic AI solution that may not be perfect.” I mean, just two weeks ago, I think it was, Grasshopper Bank introduced a brand-new small business solution. I'm going, holy crap, in certain ways, it's here. But they were very clear that they're willing to take chances to say, “We may not have the answer complete, but we're going to include the customer in that decision.”

Jim Marous (33:20):

Agentic AI may be great, but I think the first wave's going to be, “Here's something we found, let us know if you want us to complete that transaction or that event or that transfer,” whatever it may be.

Jim Marous (33:33):

But I think it's very interesting that nobody's way far ahead within the masses. There's many in the middle section of asset sizes that aren't there yet. But you are finding with your research, both on the talent and the outcomes, that the answers as to how to get to be one of the most dynamic, most progressive AI organizations are getting clear. The answer book is a lot clearer than it was two years ago when we last met.

Jim Marous (34:02):

So, I'm going to put you in a situation you're now going to be in charge of or consulting to which I know your organization doesn't do — but consulting to a midsize, let's say $10 billion of assets. What are two to three things that this type of organization must do today? Because they can't just sit by the wayside saying, “Not me,” because there are solutions out there. What do you suggest to them?

Alexandra Mousavizadeh (34:30):

Thank you, Jim. That's a great question. Maybe just before diving into answering the question, just to reiterate this point that you were flagging that the leading banks, they're out there doubling down and pulling really far ahead of the lagging banks. And as I said before, we're seeing that gap grow between the leading and the lagging banks faster and faster and faster.

Alexandra Mousavizadeh (34:57):

So, you've got to ask yourself at what point in time can you not catch up any longer? Because if you are up against banks that have poured a lot of money, focus, resource, operating models for 5, 6, 7, 8 years at some point, then there's a lot of efficiency gains that they will have made, their cost to delivery will have come down, their revenue uplift from AI will be up there.

Alexandra Mousavizadeh (35:30):

So, it's going to be really hard for you to compete as a bank unless you have circumstances around in your market, that means that you are ringfenced for whatever reason is going to be really hard to stay in business. So, I do think it is existential if you don't focus on this.

Alexandra Mousavizadeh (35:48):

So, to answer your question; if you're about to focus on this, what should you focus on? And I would say you've got to focus on everything. You've got to get every single bit of-

Jim Marous (35:58):

No relief.

Alexandra Mousavizadeh (36:00):

No, no, there's no rest for the weary, and definitely don't think you're going to get any sleep because often there's a question like what are the three most important things? And that's hard to answer because actually, it's an ecosystem and you've got to get all of the components right, and then you've got to keep leading it, and you've got to lead it hard.

Alexandra Mousavizadeh (36:20):

And as a CEO of a bank, you've got to be living, breathing every night, every weekend, every day. What is possible in your business? And are you pointing every single soul in your business to thinking about AI? Are you aligning your heads of lines of business bonuses against AI results? Are you thinking about the best orchestration where you are centralizing certain things, but you're federating other things? Can you build a hub and spoke system for your bank that allows for nimbleness?

Alexandra Mousavizadeh (37:03):

These are things that are lessons that we are seeing off the leading banks, that you need a hub and spoke system. You need to have applied AI researchers. Whether you are a big bank or small bank, you absolutely have to have that. You have to have your talent pool, you have to have your innovation levers. You need to partner with the ecosystem. You need to go to the universities and say, “I got this problem, let's sponsor a PhD program for you to solve that.”

Alexandra Mousavizadeh (37:29):

“I need to attract that AI talent, I need to be out at the academic conferences, I need to be shouting from the rooftops that this is what we're doing, because otherwise, my shareholders are going to start questioning what I'm doing, and you don't want to lose that confidence.”

Alexandra Mousavizadeh (37:45):

You also need to attract that talent. And then you need to sort of obviously be quite visible around that your governance structures are solid, so as you're putting use cases through to production, you're doing it in a way that doesn't sort of tip the applecart.

Alexandra Mousavizadeh (37:57):

So, you've got to do everything. There's not one bit that you can't do, you can't have great governance and no talent. You can't have all talent and no innovation. But you just have to have all the bits in the mosaic before it makes up the picture. And so, my advice is, it's everything. You got to focus on that.

Alexandra Mousavizadeh (38:20):

And honestly, that's what we're seeing. Even the leading banks that are far ahead, this is what they're doing. And the banks that are fast followers are saying, “Oh, we better catch up. This is what they're doing.”

Jim Marous (38:32):

So, there's laggards in your top 50, there's organizations that you can see report after report after report aren't there.

Alexandra Mousavizadeh (38:42):

Yes.

Jim Marous (38:43):

If you were to name one thing, in your gut, again, because sometimes it doesn't come out in paper as much as you peeling back the layers and saying, “Wait, I'm seeing something here that seems to be consistent from a hurdle standpoint, from a not getting …” What is that one thing?

Alexandra Mousavizadeh (39:00):

Well, I would say what we've seen is that the banks that have been sitting lower in the index and would admittedly say, “Yeah, we haven't really focused on it.” The minute that they pick the sort of the one, two, or three super strong great hires, you get the right top talent in, that's going to drag everyone with them. That then attracts further talent. It inspires the teams to think in ways that they haven't. And that can be done with not that many resources. Get the top team in-

Jim Marous (39:39):

Mindset, yep.

Alexandra Mousavizadeh (39:40):

Go find them, bring them in, say we're ready to change, tell us what to do, and let that flourish, then a lot of things will fall into place. And as a CEO, you got to be comfortable and happy with saying, “This is what we're doing now. This is why we're hiring these people, we're doing a step change.” So, I think that that is important.

Jim Marous (40:09):

Look what the leaders are doing in the world today. Look how people are being stolen from organization … now, mind you, this is a different level, this is a different industry, it's the overall think tank. But what's happening is people realize at the end of the day, they're not hiring the best data scientists, they're hiring the best implementers. So, people are actually going to move forward.

Jim Marous (40:31):

And as it is with any type of innovation, it's only going to happen if the leadership and culture align. You're only going to hire the best talent if you have leadership and culture in place. And for lack of a cliche term, if you know what your North Star is going to be, if you really know what you're trying to achieve, even if it's in a composable solution, even if it is one part of the whole data set you want to work with — I think it's important because your studies continue to show that not everybody's got it. Nobody's got an insurmountable lead when it comes to the customer facing components.

Jim Marous (41:08):

So, as far as the competitive scenario, I think we've seen that there's a ton of opportunity in the revenue side, which people are just scraping the surfaces of. And we didn't get into it too much here because I think it's very clear, not enough organizations are measuring the results in a cognitive way that can move the needle from somebody going, “Oh, by the way, we got to do more of this. They're moving more cautiously because they haven't measured it.”

[Music Playing]

Jim Marous (41:36):

And I think if you don't know what you're going to measure on the front end, you're never going to measure it later because you don't know what to measure. So, you have to know your North Star.

Jim Marous (41:44):

I think overall, for those who are listening to the podcast today, you have to look at the future banking in an AI-driven world in a multiple context. You need to maybe take on the podcast or the event I did in Abu Dhabi where I unpacked what's next for financial services in an AI-driven world. And we always have a lot of podcasts on what's it going to look like in the future of a banking transformed mindset and in leadership to make it go forward.

Jim Marous (42:16):

Alexandra, thank you so much for your time today. You again brought a ton of insights. It won't be this long again because you keep on coming out with more research that I think reinforcing what we see and what we feel, that there's a ton of potential, ton of opportunities, but we're kind of sometimes held back by our own legacy thinking on, do we really want to dip our toe in that water?

Jim Marous (42:40):

The risk is not that great in context. We have more than enough excuse, we can always blame it on regulation and compliance. The reality is those steps, we're not there yet, we're still doing baby steps. Do the baby steps, at least move forward, because otherwise, as you referenced, you're moving backwards. Thank you so much, Alexandra, appreciate it.

Alexandra Mousavizadeh (43:01):

Thank you for having me, Jim. Thank you so much.