EP 353 – Is the Future of Finance Being Redefined by AI? – Nimish Panchmatia – Chief Data and Transformation Officer at DBS Bank

Read the best-effort transcript below (This technology is still not as good as they say it is…):

Michael Waitze 0:04
This is Michael Waitze, and welcome back to the Asia Tech Podcast. We are joined today by Nimish Panchmatia, the Chief Data and Transformation Officer at DBS Bank. Nimish, it’s so great to have you on the show. How are you doing today?

Nimish Panchmatia 0:22
I am doing fine, fine, Michael, and thanks for having me on this. I’m quite looking forward to it. Yeah, it’s my complete pleasure.

Michael Waitze 0:27
I’ve been looking forward to this as well. And thank you so much for all of your flexibility around scheduling and timing and stuff like that. I really appreciate it. You’re welcome. Okay, let’s give our listeners a little bit of your background, just for some context, and then we’re going to get into the meat of this conversation, which is going to be a lot about artificial intelligence and machine learning, go

Nimish Panchmatia 0:43
ahead, right? So chief transformation and data officer means quite a lot of things. I’ve been in the role for three years. I’ve been part and parcel of the DBS transformation journey, right from the beginning, when we started on this journey, 1213, years ago, actually 15 years ago. And, you know, we started off. I mean, I started off in the organization in channel servicing, and that was things like ATMs and all that. Then I went to run consumer ops for our biggest franchise, which is here in Singapore. After that, I was the Chief Information Officer running technology and operations in Hong Kong in China. Did that for three years out of Hong Kong, and then spent three years running the frontline business, which is the corporate product, and then came back three years ago to take on this job. So I’ve done pretty much, pretty much everything from the front to the back, in the back, as you can imagine, I

Michael Waitze 1:39
feel like you and I have had similar careers, just on different services, like I was all in the back office to the middle office, then into the front office, and always using technology to make my job, no matter where I was, like, more productive. I feel like I was doing digital transformation the whole time I was at Morgan Stanley and Goldman Sachs. But that’s a story for another time. I feel like AI, we were talking about this a little bit before, when we were offline. I think AI is beyond the hype phase right now. And I’m just curious about how we can talk about DBS in particular, but let’s talk about the finance industry as a whole. How can the finance industry you think, if you look at all the stuff you’ve been through, all the digital transformation and all the lines of business that you’ve looked at, how do you capture this value of AI and machine learning as you go, as you can go through this continual process of adding tech into all the places where business operates.

Nimish Panchmatia 2:29
So I think it needs to be looked at in two parts, right? And the first part is traditional AI, what we call traditional now, which was pretty cutting edge not so long ago before, you know, generative AI term. And so for, for for the traditional AI, I think, you know, a lot of organizations have made progress. I think we’ve made significant progress ourselves. And you know, this year, for example, 2024 we’re going to attribute over 707 $80 million in incremental economic value purely from AI. And this is, you know, not, you know, applied AI somewhere here or there. Actually, this is, you know, on a test and control basis, so pretty strongly validated numbers. This is double from last year, and it’s double from the year before. So we’re quite ambitious. We’ve invested quite a lot. We started on this journey a long time ago, way back in 2018 and we determined that, you know, the next wave of digital transformation is going to come from data, and therefore invested in, you know, in people, in processes and in technology and platforms to enable us to not just do data in pockets, but right across the organization. So we made a lot of progress in that. So I think on one side, you know, using structured data, it’s, it’s, it’s become easier to attribute value from it on generative, AI, though, I think the jury is still out. There’s a lot of hype about it, and some of it is real, for sure, and we’re seeing some some significant value coming out of it. But you know, if you were to listen to some of the bigger consulting companies and the numbers they’re throwing around. I think there’s a long way before you know we we believe that those numbers will be true, if at all.

Michael Waitze 4:26
So can I ask you this, a lot of the organizations that I talked to are really tentative about even employing traditional AI, right? And now you’re already talking about, we’ve done that. We’ve been doing it since 2018 and maybe even before that. And now you’re thinking like, how do we employ this generative AI, which is a different animal, per se, right? Because the inputs are one thing, but the outputs can be really creative or really different. I use it almost every day in the stuff that I do, right? So I’m kind of familiar with the way it works, whether it’s image generation, text generation, or just like an amalgamation of search and things like that. Yeah. What is it about DBS that makes it so forward looking, right? That allows at least these tests to take place? I mean, you said you used a number. I want to make I wrote it down. I think 700 and something million dollars of incremental value. That’s a lot, and doubled from the previous year, and doubled that from the previous year. Where does this attitude come from, that it’s okay to experiment?

Nimish Panchmatia 5:21
Well, so that goes back to our overall digital transformation. And so we are probably now, if you like, on the third wave of our digital transformation. Okay, the first was really getting our tech stack, and, you know, the end to end processing done, which was a journey we embarked on in 2012 and I mean through 2020 you know, through to 2016 2017 data, structured data, was 2018 onwards. And now 2023 2024 is the third wave, which is around the generative AI. We are very, very clear. I mean, we won several awards across, you know, many, many years, being the most, you know, transformative bank, and being the best bank in the best bank in the world, and all that kind of stuff. And this really, you know, we’ve got three core components of how we approach any form of transformation. First is, as I mentioned earlier on, being digital to the core. I mean, a lot of people apply digital lipstick. These are the people that you mentioned who are tentative, yeah, in terms of using it, because they haven’t really digitized to the core, and therefore they don’t have the kind of instrumentation to be able to pull out all the data or even generate the data. Forget about pulling it out, generate the data to then do analytics on it. So that’s the first part, right? The second part is we are completely customer obsessed, right? And in this we mean that everything that we we want to do, we look at, we think that has some legs in it. Starts with, what does it mean for the customer? And then we work backwards, right? And so if it’s not a if it’s not so, if it’s not a value to the customer, we won’t bother pursuing it, right? And how do we establish value to the customer? We actually go and talk to them, right? Which may sound pretty straightforward, but you’ll be amazed how many people come up with products, throw them out and realize, oh, customers didn’t want it, right? And so we talked to the customers up front. We prototype up front, right? And this gives us this experimentation culture, which leads me to the third core component of our transformation DNA, if you like, which is, you know, having a startup culture, right? And then banks don’t talk about being startup, but our startup we, you know, we’ve been pushing this through the organization, and we’ve been very clear what it means to be a start, you know, to be a startup for us, one experimentation, we actually celebrate failure, right? And we’ve been doing that for years. Somebody’s got a hypothesis, they’re going to try it out. It fails. At the end of the year, all of these get put together. And, you know, the biggest ones, they get awards, right? For driving how to fail. Of course, the ones that tried and try and be successful are also rewarded for for what they do. So we’ve embedded that, right? We do hackathons, we run, you know, several experimentation with with multiple partners, not necessarily within, within the bank, but from external. We interact with startups. We don’t think of startups or fintechs as the enemy. We actually think of them as partners. Yeah, and, you know, see how we can leverage and all that. So we, you know, experimentation is deeply embedded in our DNA, and has been for for about a decade now. So

Michael Waitze 8:35
the large language models of the stuff that’s driving a lot of this stuff in generative AI, are you as a as a team and as a group, kind of implementing your own LLM, so you’re not using the public ones, I presume you’re using very specific llms internally, giving them to people inside and ask them to innovate around that. No, no,

Nimish Panchmatia 8:55
we don’t build, we haven’t built our own. LLM, just, I don’t think we need to build our own LLM there are hundreds of 1000s of them out there. I mean, you go to hug your face, and you’ll find several 100,000 if not a million by now. So llms are available. It’s not really, the question is not really about llms, right? There’s plenty out there, right? Some you can bring on prem and secure them so that there’s no external connection. And in some cases, you got to use the public ones, because the pace of change is massive, and you bring something in, and you engineer it keeping pace with that change. With the pace of change is not keeping up with that change. It’s hard work. But anyway, back to the LLM thing. So we use external llms Mostly, and but we have a wide variety of them. So at the last count, I think we had about 38 different llms that we use depending on the use case. There is generic uses. For example, our internal GPT, or equivalent of chat GPT, we call it DBS GPT. But. That one uses a public, super large language model from one of the big techs. We have use cases around compliance and legal, which are quite specific, and then we use very proprietary llms that have been developed by smaller companies. And so actually, it’s the use case that drives, you know, the selection of the LLM, as opposed to, you know, do we? Should we create our own LLM? I think the expertise required for that is very different from the typical analytics resources that a bank would have. And there are a lot of experts out there. Let them do the that part of the good job. We’re good at running bank processes and running bank technology. And you know, how do we bring that and create an ecosystem that’s the most efficient?

Michael Waitze 10:50
How do you, and maybe I can frame this in this way, right? A lot of big universities, whether it’s in the United States, in Australia, in other places, right? Do a lot of experimentation, a lot of research and a lot of sort of, let’s develop the most cutting edge tech, but then getting that tech from the research phase into the implementation phase has ended up being hard for a lot of institutions. It doesn’t feel like that to me with DBS. And I’m just curious, like, how you get this experimentation mentality into implementation and like, is there a process around trying to decide, like, what’s going to work, what’s not going to work, and stuff like that, if you know what I mean, because if you’re and the reason why I ask is because, if you’re enabling your teammates right to be able to experiment like this, whether they’re winning or losing, or you said, both of them get rewarded, how do you then decide, oh, this thing needs to get implemented. And I know there’s a customer side to it too. So how does all that work together?

Nimish Panchmatia 11:41
So maybe I’ll take a step back and explain my role. That might answer the question better. So what we mostly talked about until now is my chief data officer role in the bank, but I also have the chief transformation Officer role. And the reason why these are together are I’ll explain. So, you know, transformation for us is a very simple, you know, process, right? We have, we take big ideas, we look around callers, we assess what’s coming three years, five years away, we will bring that in, understand it properly, engage with with the top leadership of the company, including the CEO. And when I say engage, it does not mean, you know, a 15 minute, 30 minute thing I’m talking about, you know, 10s of hours spent on a specific topic, understanding what is it, what are the implications of it? And once we’ve determined that, yes, we understand this well enough, and yes, this has implications to us, right, we will then determine, Okay, what’s the roadmap? How? What does this? What will it look like in two or three years? And therefore, what does the organization need to be prepared for? For this, not to be a surprise, once that’s determined that you know that that’s if you like, the book of work, we then take this book of work, and we break it down into small chunks, right? And these small chunks are understood by the masses, the population, the entire bank, as opposed to, you know, the top 1520, 30 people in the bank understanding it, you know, we break it down so that it’s understandable. And once people understand what it is that needs to change, then they understand, you know, from their existing processes, what’s the delta? And at this point, we encourage to say, Okay, we need to start building hypothesis. And when you build this hypothesis, you need to go and test them, right? And that’s a structured process. Now, you know, that’s experimentation at its best. Yeah, right. We do, and that’s internally. Externally, we do the same thing if we determine that, you know, there’s a product coming around the corner, and for example, you know, one sort of work around, you know, authentication, for example, right? Which move from passwords to fingerprints to face recognition and all that kind of stuff. You know, if we feel like concepts really going to take on port. We will develop prototypes. We will go to customers and create customer focus groups and spend, you know, hours and hours with them, explaining to them how it work and if it resonates. We will then go and, you know, take it from a paper prototype to kind of a, you know, quick, quick and dirty app, build it, go and test it again, right? So it’s part of the process. You know? It’s not that we stand up and say, Hey, here’s a great idea. Can you go and experiment with it? No, no, you know, it’s step by step. And you naturally lead people to a point where there is nothing else to do than to try and test and once they’ve done that, their learnings, right? The ones that fail tell you that you miss something. The ones that pass say you okay, you thought about this through probably, but the ones that fail are very important, because they’re the ones that inform you far more than the ones that succeed, all right? And that goes into your design principles.

Michael Waitze 14:56
Absolutely. I really want to know how you do this mindset thing, though. Because, look, I worked at some big companies. I worked at Morgan Stanley, I worked at Goldman Sachs. We considered ourselves pretty innovative, but I never felt like failing was an option. It definitely was not rewarded. I’ll give

Nimish Panchmatia 15:12
you that for sure. Look, I’m not saying that every failure is accepted. No, I

Michael Waitze 15:16
get that. I get that. But you understand what I mean, right? There was no failure accepted when I was at Goldman Sachs, it just wasn’t acceptable.

Nimish Panchmatia 15:22
Well, a few simple components around it really. First, you know, I mean, you’ve got to prepare people. You’ve got to communicate it in the right way. And it’s it, and there has to be a demonstration of it across the board. You know, across the organization, the CEO must say, right? And not only say, but, you know, prove it that he’s okay with it. And has to percolate all the way down, right then. And so there’s got to be the senior level buy in. Secondly, you’ve got to provide environments for people to be able to test and experiment, right? And you know, many senior leaders say, yes, it’s okay. We love experimentation. Please go and experiment. But then people say, Okay, I’ve got this idea. I don’t know where to go to experiment, right, right? So this is where, again, the transformation group comes in, right? So we have the tools and the platforms where people can experiment, including, you know, we have a 1500 square meter facility we call the innovation lab in the west coast of Singapore, and this is entirely dedicated for experimentation. So you’ve got an idea, and your team needs space, you need resources, you need tooling, you need connectivity, or you need a partner to come in and all that. You got the space go and do it. You don’t don’t have to worry about budget. You don’t have to worry about all these things, right? Third is around training, so we spent quite a significant amount of time curating and delivering training around, not necessarily experimentation, but how to take new ideas to fruition, right? And that one component is

Michael Waitze 16:58
experimentation. So a lot of people will listen to this, and they’ll think about artificial intelligence or emerging technologies, and think that this is something that, just like the tech teams are experimenting with, right? Like you said, there’s this really large facility where the testing goes on. I don’t think that that’s what you’re saying. But how do you then get this communication going between the people that are actually building the stuff and the people that are using it, and kind of cross pollinate those ideas. Do you know what I mean? So that, yes, the gal who’s sitting in the front office, who’s dealing with clients thinks, if I could do these three things and combine them into this thing, do you know what I mean? And then, but I can’t build it now. How does all that work?

Nimish Panchmatia 17:35
So two things there. First, you know, as part of our first wave of digital transformation, one of the things that we did was we reorganized product and technology and brought them together, okay? And we so we call it platform, not in the broader sense of the public definition of platform, but our internal we call it platforms. And we constructed 32 platforms, and with this concept of two in a box, right? And so that is the business people, the product people, the sales people and the technology people. They own, the build budget, the investment budget, collectively. So one person doesn’t decide it’s not the business guy going, Hey, I got this idea. You go and make it. Or the tech guy goes, Hey, I’ve built this, you go and use it, right? So it’s commercial, that’s, you know, the first part of it, right? Secondly, since 2021, we’ve extended this concept of two in a box, right across the organization, right? And so we, you know, in the in the general world, it’s called Agile at scale, right? But we’ve constructed, you know, Journey thinking through the organization, where it’s not just two in a box, but it’s N in a box. Anybody who delivers value to a customer, right? Is brought together in a in a horizontal construct, and they co own every single metric, outcome metric, right? That is, that is relevant to that customer outcome. This drives pretty, pretty good alignment, right? And, and even even to the point where even compensation is determined in that way, right? So it’s just not a hey, let’s all get together, and we’ve got a couple of KPIs that we own together. It actually is, you know, right down to, you know, the outcome determines the compensation, and the compensation is distributed based on this horizontal structure. So these two things is what allows us to drive scale in terms of experimentation. It’s

Michael Waitze 19:38
a really interesting structural solution to what, in some places, could be an intractable problem, right? In other words, you don’t encourage people to get together and do this. You just put them together and they end up being on the same team. Again, we did something like this very similarly before, and it worked actually really well. You don’t have to create a separation of responsibilities. You create this continuity of responsibilities, and everybody. Running in the same direction. Do you feel like this testing creates an internal sandbox as well? Do you know what I mean? So if you are testing things that the implementation of a failure, and I put that in quotes, right, is so small that the learnings end up being way larger than whatever the potential failure is

Nimish Panchmatia 20:18
in I think in my in my case, I would say, if not all, very nearly all cases, yes, I like that. The learnings are far bigger than the failure. Yeah, yeah. So do you want a quick example? Yeah, I

Michael Waitze 20:31
do. I was just gonna say, if you can

Nimish Panchmatia 20:33
example, so, you know, we, you know, we launched this service, and it was around getting kids to open blanket counseling to their parents. And you know, we were finding that there was a drop off rate of 60 70% and so, you know, people thought, oh, okay, this is a bit of a problem. It shouldn’t be 60 70% it should be 5% and so there was some hypothesis. People said, Okay. And the first hypothesis was, oh, it’s going across five different pages, and it’s a 90 click process, and that’s what people are bulking off at, right? So hypothesis one, and they said, Okay, well, you know, should we should we experiment, right? And let’s do a customer discovery and then do an experiment. So okay, they went and spoke to customers, and if got a couple of insights, and customer was like, Yeah, I’m not that bothered about it, but there’s something wrong with the upload function, right? And the upload function was coming at about 70% of Journey completion, right? So the guys came back and said, Okay, well, it’s not the number of pages, at least. That’s not what this set of customers said, Right, right? What they said was, oh, it was the upload thing. That was the bulkhead. And so, you know, rerun the process and discovered that more often than not, the connectivity between the upload function and connecting with an external partner, right, was a bit of a problem. And so, you know, went back to the partner, we fixed the problem, and, you know, gave dedicated bandwidth and set it out there, and block rates dropped to nearly zero. Now, if we hadn’t done this right, we would have gone and recreated the entire web page, tried to reduce this from, you know, 96 steps, or whatever, until four, something like that. Yeah, to four, spending significant tech resources in the back end, right, right, to get it done, and we would still have that same problem. So, you know, there you go.

Michael Waitze 22:33
But there’s a really interesting mentality thing going on here, right? Because, like, I always inside my own business, I always say, don’t make me ask the next question, right? And what you’re doing is you’re you’re encouraging people to follow up. You say, Okay, you said 96 steps. But maybe that’s not the thing that’s actually doing it. Let’s dig a little bit deeper here and find out why, because maybe those steps are necessary. But maybe the thing is that whether it’s the upload problem or some other issue that’s taking place, if you can identify it, you save all this extra time on the back end. Yeah, and I do think this as well. And tell me if you think I’m wrong, all you really need is one person to create like, one success like that, and then from then on, they’re willing to take way more risk in their implementations and participate more in that sandbox experience with like. I have another idea. Can we test this one out as well? No, oh,

Nimish Panchmatia 23:24
absolutely. And, you know, and when, when you celebrate that across the organization, right over 40,000 people in our organization now, when you celebrate that many 1000s see it, right, yeah. And then they know there is a sandbox, they know there is a place to go. And they know, wow, okay, I don’t need to be fearful. Because, I mean, when you tell somebody you’re going to fail, it’s the first thing you feel is fear. Yeah, right. So you know, if you provide that environment that allows for this, and actually, you define success differently. You don’t define success as whether it did you make money or not make money? Did you learn something that you didn’t know before? Completely shifts the pandemic so

Michael Waitze 24:09
that that’s another really interesting question, right? Because at the beginning, you mentioned this idea that, like some of the stuff they’ve implemented has created, like $780 million of incremental value. It’s a lot, but what other ways do you measure these, whether it’s aI driven initiatives, or any other sort of transformational initiatives, to make sure they’re actually having an impact? Besides me, I mean, I guess NPS, the Net Promoter Score, probably matters at some level. But how else do you measure this stuff so that people actually know, like that thing that I did actually worked?

Nimish Panchmatia 24:36
So we’ve got four key metrics, and the first one we always start with is around the customer Yep, right. And this is not around revenue. This is around customer toil, customer satisfaction and customer experience. And you know, maybe on another day, we’ll talk about the difference between experience and satisfaction. But you know, and we. Then that comes E, N, P S, and sorry, N, P S and all that. So that’s the first one that we measure. The second we measure is employee, right? So any process or any value that we deliver to the customer, what does this mean to the employee, from an end to end perspective, how much toil is there in the process? Right? And how can we remove this soil? How much information search is there? And how can we reduce that? How can we make sure that the workbench they’re working from is equipped with all the relevant information for them to be able to make a decision straight away? And how much of this can we digitize such that there is no paper pushing and manual processes in the middle. So that’s the second part. The third part we measure is around risk, right? And so whatever you do, right? You got to ensure that you don’t create any risk for the customer, maybe cyber or fraud or whatever the case may be, but also for the bank. You know, what’s you know, if you’re going to add an additional process, you got to make sure that you don’t add any additional operational risk, and as a result, have some failure. And then the fourth one, as we spoke about, is value. That’s the dollar number, it’s revenue, it’s cost. That’s where the 780 million comes from. Can

Michael Waitze 26:13
we dig a little bit deeper into the into one of the things that you said, because this is also really interesting to me, and not a lot of business leaders talk about this, and that is this idea of like having initiatives that take this emerging tech, whether it’s AI or any other tech, to make the employee experience better as well. Do you know what I mean? Because we spent a lot of time talking about being customer first, which I completely understand, right? You want your customers to be happy and satisfied and have their satisfaction experience be great. And again, you’re right. We can spend hours talking about that, but if I call a company, or if I’m working at a company and I’m not satisfied in my job, or if it’s too what’s the right word, if it’s too hard for me to get my stuff done, it’s gonna filter out into what that customer experience is like as well. Do you have special initiatives that do that internally to to make just everybody’s job easier?

Nimish Panchmatia 27:01
Oh, absolutely. So, you know, we have a bank wide initiative we call Workbench workflow, which means, wherever there are handoffs, what can we do to connect? Right so we try and flow work 100% electronically. Right now, we’re not at 100% because there’s always exceptions. We will try and get and then on top of these flows, we put work benches right. And this work benches will will pipe information right for the person who’s on the who’s on the bench to make a decision, and it will type the information from various different workflows right. And maybe one of the workflows could be from risk management. For the guy who’s managing credit risk, one could be from operations, one could be from the branch. One could be from sales, you know. And so, you know, to make a decision, you need different disease, you need different information to get to a decision point. So we try and bring that together into into one. That’s one two. We actually have a metric, right, which is employees, employee hours saved, right? And so there was a process, you know, for example, a trade document, and that used to take 15 hours for the employee, for multiple employees, to fulfill from end to end, right? And you’ve got a million of those a year. Yeah, right. So that’s 15 million employee hours. Now we automate a part of it, right? Or we stitch together through Workbench workflow, a part of it, and that removes a million hours. And so we measure that, and it’s extremely powerful, yeah, right. And when we go back and we say, Look, this year, we save 10 million employees our employee hours, right? We do the same thing, by the way, for customers as well, right, right, in terms of customer hours saved, right? And so employee hours saved is a major metric. We also measure enps, so we have an internal equivalent of NPS for employees, employee net promoters. And I mean, as of last count, in our horizontal structure, which is about 60 65% of the bank the enps is 68 people love working like this. I was gonna say, you know, I mean, most organizations, 100, 150, years now, have always thought about it as pillars. You know, you got your vertical structures, and work moves from one structure to the other structure. Actually, customer value is not delivered that way. It’s delivered horizontally. And when you put put people this, like people together to do it that way, you know, everybody goes, Oh, no, it’s going to be very difficult to work with. But once it’s bedded in. I mean, you know, 68 is more than world class.

Michael Waitze 29:40
Yeah, it’s way more the world class. I like to make analogies, and I like to think of like to really simplify stuff. You can have the greatest chef in the world making the greatest food, but if the waiter or the waitress is bad, yeah, what I mean though, like it’s the same thing, and those are definitely different pillars and different. Silos, right? Because the waiter, and waiter is not going to go in and cook the food, and the chef is not going to serve it. And I know I’m simplifying to make a point, but again, from a customer satisfaction and from a customer experience standpoint, that’s a horrible experience. If the food’s amazing, sorry, go ahead, yes,

Nimish Panchmatia 30:14
exactly. I completely agree with you. And actually, this is one of the reasons we spend a lot of time around, you know, empathy, and, you know, formally training people for to be empathetic. And it’s not easy. It’s very easy for you to make a statement, right? Yeah, oh, Michael, you’ve got to have empathy on that. I’ll tell you some story. You got to have empathy on that, right, right? And, you know, intuitively, you know that, you know, yeah, you know, poor guy broke his leg. I’ll, you know, I’ll be, I’ll be empathetic to him. But there are so many situations in life that are not black and white, the gray right and got determined the right. At what time do you get the display the right level of empathy? Because if you over display, it’s not authentic. You under Display, you know, like your waiter analogy, you you you know, you lose the customer. And so it’s really important to get that balance

Michael Waitze 31:08
right. It really is, do you can we talk a little bit about startups, particularly in the context of artificial intelligence, like I’m presuming, now that I’ve heard all of this conversation with you, that there must be some kind of process internally to understand with whom you’re working, how you engage with them, how you then integrate the products into there, and then how you decide, like, we’re just not going to do this, or we’re going to put it into the sandbox and take it further. Is that, does that? Is that part of your remit as well? And I presume there’s some structure around that. What does that look like?

Nimish Panchmatia 31:38
Yes, we do. And you know, we have a whole accelerator program that goes from outreach to, you know, identification, but we do it in a slightly different way from other people. Most accelerated programs are seen as, you know, there’s an outreach and says, Okay, look, we’re so and so company. Give us, you know, give us your information and your pitches and your pitches and all that kind of stuff, and we’ll determine whether we’ll put you in the accelerator or not. We actually do it slightly different. We start with going to our business leaders, and we ask them for a problem statement. We then publish these problem statements and we ask startups to say, hey, you know, do you think you can solve this problem? And then we’ll bring them in, and, you know, we’ll work through that, and of course, you know, through the sandbox, and then all kind of stuff. But you know, we determine success metrics right at the beginning. And success metrics do not necessarily mean go live with production. Some of these are, you know, you know, you get to a point where you’ve learned enough that you don’t need to do more, so, yeah, so, you know, it’s a very structured process, and we’ve been through hundreds and hundreds of startups over the years. You know, more and more, you know, we’re finding that our internal culture has changed to the to the point where there’s a lot of ideation that’s happening organically. And therefore, you know, unless something novel comes up, we’re not, you know, we don’t bring them at that scale. Or, you know, more interested now, I would say, at a point where we talk more to academia as well, particularly around AI, because, you know, when, when open AI, you know, was revealed to the world. It came as a surprise to many people. I mean, we were, we were dabbling around with llms and experimenting and all that, but it was a very technical nature. We hadn’t really applied business, you know, business lens to it. So, you know, with that, we thought, okay, you know, maybe it’s time we start talking about, you know, not what is best practices and what people are already doing, but what are the next practices? Right? Right? What’s coming in? Six, 912, 18 months. I think anything more than 18 months nowadays is too far

Michael Waitze 33:54
out, right? Gambling. Yeah. You really have no idea.

Nimish Panchmatia 33:56
Yeah. So we spend time with academia and all that. And then, you know, there are great forums out there that we attend, and we work together with partners. You know, we have our own lab, of course, but we worked close, quite closely, with other companies.

Michael Waitze 34:15
Do you interact with academia globally? So are you constantly Sorry? Go ahead.

Nimish Panchmatia 34:21
Yes, we do. In fact, quite recently. I mean, in absolutely, for AI, we just did a case study that Harvard Business School will teach in there, in MBA. So we released, we had a press statement on that few weeks ago, which we can share with you after this. So we work with them. There’s a few more coming over the next Actually, there’s one probably going to be released next week with another big business school. So we work with that. We work with the likes of MIT on specific topics. Same with Stanford, of course, at home. Very importantly, we work also with NTU and. SMU, so yes, we do,

Michael Waitze 35:02
yeah, really interesting stuff. So, you know, you just mentioned this idea of, like, what’s gonna happen the next six months, eight months, 12 months, or whatever. It’s hard to figure that out, right? But it is. This stuff is rapidly evolving, particularly in the AI space, right? When you look at what the bank’s looking at, because from your perch, you’re seeing a ton of stuff, right? Do you have, like, a couple of places where you’re definitely focused on going forward? Like, a lot of innovation has been done already, a lot of implementation has been done already. But when you’re looking from your perspective, do you look and think we need to do more here or do more there kind of thing?

Nimish Panchmatia 35:35
I think, look, the tech is evolving really quick. Yeah. And so one of the things that’s really important is to be current. Because, you know, you might look at something today, find it very interesting four months later, it’s redundant, right? So you’ve got to stay current, right? So that, so that’s a really important point. But, you know, I think that I for you’re gonna do the question again. Sorry,

Michael Waitze 36:07
no, it’s okay. So my question really is around this it’s like you have this really large remit. People are constantly coming to you with ideas, or coming to the team with ideas, but they’re really high level places where you think we need to focus on because here’s the thing, your customers, whether they’re business customers, SME customers, right? Or even individual customers, are constantly interacting with AI large language models on their own, right? So if they’re individuals, and they’re writing stuff, they’re doing, they’re using LLM, so using all this technology on their own, and then they’re coming to their their banks or their other companies and saying, Hey, how come that doesn’t exist yet. I can do it that way somewhere else or at home. Are you seeing those things like, where do you think that focus has to be just based on customer expectations of what they think is going to happen? Does that make sense? Sorry, so

Nimish Panchmatia 36:58
yeah, I lost my channel. Yes. Completely makes sense. So look, that was the point I was trying to make. Look, this thing’s evolving really quick, right? What is really exciting today in four months is redundant, and customers are at the forefront of this. They are experiencing all this new technology, not from banks, but from all the other places where they and banks, is probably one of the smallest interactions they have, maybe. And so they keep asking this question. So it’s very important to be current and in terms of what’s going on and understand again, coming back to our process, right? In the past, looking round corners was a three year exercise. Now, looking around corners is a six month exercise, right? So you got to keep and the moment you spot something, you’ve got to bring it in, right? And you figure out, what does this really mean for us? And figure it out quickly, right? Invest the time, and then break it down into programs and make sure that people understand what it is and got it so it’s an intuition. It’s not a one off exercise. And say, okay, look, here are the five things we need to do. It’s continuous second, you know. And it’s really, really important, as this evolves, the technology is moving into fast pace, regulation is going to catch up. This is going to, you know, and so you got to be very careful about that component as well. And, you know, AI regulation is not very mature as such, right? I mean, a lot of progress has been Yeah, a lot of progress has been made, right? Particularly, you know, with forward looking regulators like mas, a lot of progress has been made, but in many components, it’s still not quite there, right, because it’s evolving so quickly, right? So that’s going to be taken into into account as well. But, you know, I mean, I’ve seen use cases in AI, from health industry to manufacturing, you know, I’ve seen stuff in retail, which I look at and think, Wow, what a wonderful idea. But that will never fly in a bank, right? Simply because, you know, it’s a regulated industry, plus, also, you’re talking about people’s money and savings, right? And life. So, yeah, can I ask

Michael Waitze 39:03
you this and then maybe I’ll let you go? I don’t really want to let you go, actually, but I think that we maybe have you back and just have a conversation about another kind of vertical bit of this, if you don’t mind. Yeah, but I’m really curious, because you’ve worked in the entire region. I think you also said you worked in London as well. But I feel like there are some really big benefits of working in a place like Singapore and having really forward looking regulators, particularly, and just a forward looking government infrastructure as well, particularly as technology is rapidly changing, right? So you’re not in a place where the sort of infrastructure of the city itself is getting in the way of the innovation. And I’m just curious about what you think about Singapore itself as a FinTech Innovation Hub, both for banks and short tech like all that kind of stuff. Do you feel like it’s a real benefit to institutions like DBS?

Nimish Panchmatia 39:53
Oh, absolutely. I mean, you know the pace at which, you know the the C. System works is actually in sometimes even more forward than than industry, right? And I’m not talking about the financial industry here, yeah. And you know, it’s, it’s a, it’s a great place to be. And, you know, there’s this space for experimentation, space for incubation, and when you’ve got great ideas. There’s a lot of support from the system. You know, there are, you know, there have been grants around AI and all that, which for, I mean, we don’t need the grants, but for smaller companies, this is really, really helpful, you know, you know, nurturing the community, you know, the innovation community. This has been going on for four years. I think, you know, it’s what really helps is it’s not a question of, I need to protect what I’ve got, right? It’s more about I know what I’ve got. How can I grow this in the safest, you know, possible way, and most ethical way? And that drives a growth mentality, as opposed to a protectionist mentality, right? And so that’s, that’s what makes it amazing. Okay?

Michael Waitze 41:04
I think I’m gonna leave, I think I’m gonna stop there, because it’s a great way to kind of explain just the ethos of Singapore. It’s like we’re not getting in the way. We’re just giving you the tools and the infrastructure to create growth and not to protect the stuff you already have. It’s a great way to do this.Nimish Panchmatia, the Chief Data and Transformation Officer at DBS this was awesome. And please, you have to come back. I’m saying it while we’re still recording, because I really enjoyed this. I hope

Nimish Panchmatia 41:26
I really enjoyed the chat, chat, Michael, and I think, you know, it’s great because it’s more a conversation rather than a you know, but I learned a lot too, yes, yes. And, you know, I look forward to doing it again with you on a on another topic. Thank you. If it’s of interest, of course.

Michael Waitze 41:41
Oh, definitely it’s of interest to me. Thank you so much.