Podcast: AI Distribution for 5000+ Banks, with Fiserv Co-Head of Financial Solutions Srini Krish

Thanks for your time and attention,

Matt & Lex 🙌

Lex Sokolin:
Hi, everybody, and welcome to today's conversation. We are privileged to have with us today, Srini Krish, who is the Chief Information and Operations Officer (recently promoted to Co-Head of Financial Solutions) at Fiserv. Fiserv is one of the largest and most fundamental companies in financial technology and has launched a variety of artificial intelligence partnerships and features recently and so. I'm really excited to talk to Srini about those and about his career. So, with that, welcome to the podcast. Srini. How are you doing?

Srini Krish:
Lex, thank you for having me. It's a pleasure to be on your podcast, doing well and good to be here.

Lex Sokolin:
Fantastic. So, there's a lot to go deep into, but we're just going to do some definitions. What is Fiserv? What does Fiserv do. And then what does a chief information officer do.

Srini Krish:
Well so first let's start with Fiserv. Fiserv is very unique in the sense that it's one of the original fintechs. Before fintech became somewhat fashionable. We've been in the business for close to five decades. We're probably one of the only ones in the world that sits at the intersection of commerce and banking. We provide technology solutions to both the merchants, if you will, but from a merchant standpoint. Small, medium, large, brick and mortar and e-commerce merchants. And similarly on the banking side, we provide technology and solutions to some of the small, medium and large global banks in the world and community banks and credit unions in the US. So, we again, you know, we have both sets of clients, and we are not only in the US, but we also have an international presence in India and in APAC. As to what a CIO does, my current responsibility includes both technology and operations. From that standpoint, I work very closely with the product organization to execute to our product roadmaps, working closely with our clients as it pertains to delivery and the services that we provide in terms of keeping our systems and our solutions up and running.

Lex Sokolin:
Amazing. So, over the last 20 years, you've had a pretty broad mandate across large financial institutions, from JPMorgan to First Data to Fiserv across operations, technology, it. Can you give us an outline of that journey? More than anything, I'm actually interested in how the environment has changed. Like the outlines and the remit of a technology facing role has changed quite profoundly. So, tell us about that evolution.

Srini Krish:
You know, sort of taking myself back in time here. You know, I was born and raised in India. Grew up in India. You know, I went to school for engineering. So, I'm an engineer by trade. You know, came to the United States in mid 90s and started off, like most people, programming in mainframe technologies and COBOL and six, so some of my consulting journey in late 90s, working with state pension funds, delivering solutions for pension systems like police and fire and state data employees and so forth. Somewhere along the line, I started working with JPMorgan Chase as head of technology for some of their consumer facing departments, where I spent close to ten years.

Lex Sokolin:
I think there are two threads I want to pull again in that 20, 30-year evolution. One thread is just what it means to be doing technology and finance. And you touch this a little bit. You said fintech before it was cool. And first, I'd love to know what you mean by that. There definitely was a moment in the 20 tens when fintech sort of before blockchain and AI, was the theme du jour, and that's had a bunch of implications for how the roles are perceived. You know, who has power in organizations and where the capital goes. But yeah, maybe if you could talk about what it meant to be running technology at JPM in the 2000 versus what it means to be doing fintech in the 20 tens versus what it is now, when you're sort of this like financial lab.

Srini Krish:
Yeah. No, it's a great question. I mean, if you go back to, you know, late 90s into 2000, it was the .com, right. I mean, Chase.com was, you know, conceived and implemented very, very early on in the early 2000. And then came the Chase mobile. Again, if you really think about what a true sort of financial technology is, is that is the intersection of people being able to access their money and be able to move their money in a very sort of safe, secure and a frictionless way. And that's essentially what a fintech does. Right. And, you know, we are the sort of the technology arm of a lot of financial institutions in the US that doesn't necessarily have the wherewithal, like the likes of JPMorgan Chase and or Wells Fargo or Bank of America, where we end up being we as an officer end up being, you know, the technology extension to the community banks and to the credit unions in the United States. So, we pretty much, you know, provide soup to nuts from a financial services perspective, everything that a bank and their customers would need. Right. It starts from, you know, the core, which is the general ledger behind that takes deposits, you know, processes, checks and send statements and so on and so forth to credit cards and debit card platforms that authorizes in, you know, processes, transactions to digital and mobile banking for these banks where their customers are able to log in online and or use their mobile applications to access their funds.

Lex Sokolin:
Got it. And just in terms of the amount of capital that you've seen come into the space, you know, I imagine it's been like a rush of change in terms of venture and then growth equity and then SPAC and digital asset treasury investment into financial technology. I remember back in 2012, I was looking at the percent of venture investment that was going to fintech. I mean, the numbers were pretty embarrassing. I think less than 2% or something like that, where financial services is 20, 25% of the economy. And now we have kind of enormous chunks of money going into the sector. How has that changed again, like over your career? Has it been easier to get capital for innovation or like what's the dynamic around what you need to do to unlock growth and experimentation?

Srini Krish:
Yeah. I mean, like, you know, it really comes back down to the fact that the addressable market continues to expand, right? Even if you think about, you know, the unbanked or the underbanked segment too. You know, today we refer to it as embedded finance. Right. So, for example, some of these, you know, companies that are involved in gig economy, are trying to be able to provide services that are intuitive from a quote unquote financial services standpoint, even within their own ecosystems. Right. So, you know, if you think about folks that are working part time, either driving cars and or delivering goods they don't generally tend to think about, you know, banking in their day-to-day sort of, you know, going about doing their jobs. Whereas those companies have started to think about how to provide those services. Think of an Uber driver, for example, that is driving around their cars and, you know, pulls into a quote unquote automobile, you know, mechanic shopping. You know, in this case, providing them with some on the spot, you know, maybe loan for them to be able to get their tyres replaced or get the car serviced and things of that nature wasn't necessarily a use case that we would have thought about.

Lex Sokolin:
So, we've gone from a world where you provide technology to a bank, like you create a banking corps and there's a bunch of banks that sit on it, right. And then you might have a payment processor and the networks connects to the processor, and merchant connects to the processor. We've gone from that world to finance is embedded into everything, and it's kind of bleeds out into the long tail of human experience. And commerce is now everywhere. And so, commerce is tied to payments. It's tied to banking. And you're kind of extending these systems much further, and I would argue much further than they were really designed to, to extend into, which takes us to a couple of really interesting announcements from Fiserv around partnership with OpenAI and the launch of agent OS. Can you tell us about what those are? And then I'd love to spend some time digging it apart.

Srini Krish:
Yeah, I would love to. Before we get off the previous subject, just to not lose sight of the fact that, you know, as you are very well-articulated in banking and banking institutions and credit unions having cores and the processors having connections into the issuers and to the merchants. That's still very much, you know, bread and butter for us, right? So, I don't want to sort of not forget that for that matter. The world continues to evolve. You know, the market continues to expand into use cases that we talked about. So, it's interesting to sort of, you know, continue to pay attention to the bread, bread and butter at the same time.

You know, we continue to expand and push these boundaries into new use cases and new places. So, it's always good to sort of not forget that, but just to quickly switch topics on to your question, if I were to sort of like step back for a second. Right. I mean, again, this whole AI in the last 18 months has been interesting in terms of how fast it advances on a day-to-day basis, even on an hourly basis. But if I were to step back even before that, for that matter, we have been sort of in this path of Robotic process automation. Intelligent process automation. Machine learning. You know, these things sort of coming together for us to gather and leverage intelligence from data for us to be able to either automate processes and or make decisions in some form or shape. The AI becomes a natural extension of that journey, if you will. Right. At least that's kind of how we kind of got into AI. So, from that standpoint, you know, we always have thought about application of AI largely in I would say this is my personal characterization, but, you know, largely in four buckets.

Lex Sokolin:
That is a comprehensive answer. So let me pull on a couple of things. The first is just for our listeners to understand the shape of your customer base, I think is really important. So, you've got in the US lots of banks and credit unions that might not be the largest bank in the world. And often, you know, these institutions are focused on trust and safety and their local community, but they're not super savvy. So, you know, in my experience, it's very difficult getting them to adopt any sort of technology and software. And if they do adopt new tools, it's going to come from their core provider. What is the conversation with them like? You know, is Fiserv the saying, okay, we gotta move into this new AI world because everybody's doing it. Meaning like all, all the companies are moving there, so we have to compete and therefore we have to move our customers there. Or is there demand organic demand from these folks? And then if there is, who's it coming from? Like, is it coming from the CEO? Is it coming from the technology person? Is it coming from, you know, the 18-year-old intern who's really good with Claude code, like what's the relationship there?

Srini Krish:
It's a great question. So just to level set, you know, sort of the relationship we have with banks and credit unions in the United States, we work with roughly give or take about, you know, 5000 plus institutions. These are regional banks, large, medium size, you know, community banks as you can, as you rightly refer to. And credit unions, they are very savvy, you know, so they are digitally savvy. And again, it goes back to the customer base. Right. So, in this day and age, everybody would like to be able to check their balances on their mobile phone. Be able to make their build pays on their, you know, on their mobile phone, have digital wallets and so forth.

The motivation always comes from what can I do better for my customers. Right. So if you think about a community bank and a credit union, you know, if they can turn around a loan, you know, that's a loan application that gets submitted that, you know, typically takes anywhere between, you know, call it a few weeks to be able to process a commercial loan application to turn around and be able to provide an answer as a yes or no for a commercial loan, if it could be done using LLMs and agents in a sort of very sort of secure way where you have the appropriate controls in place, appropriate auditability in place and appropriate observability in place, and you can do that in a matter of a few hours to maybe a couple of days. That dramatically again changes, you know, the customer experience if you will. Right. So, you know, to be able to build an agent orchestration platform with that level of auditability and observability requires a significant amount of investment and upkeep. And that's kind of where a file server game plays a role, that we could do that and be able to provide that back to our customer base at scale and with the right level of controls, at the right level of comfort. And yet these banks can take advantage of it and be able to provide better customer service and or better productivity for themselves.

Lex Sokolin:
This question is a bit amorphous, and I think it's more of an industry observation. You must think of it on behalf of your clients. If I'm one of these small banks, right or not small, but like regional banks, and I rely on a Fiserv for my technology roadmap effectively, and I act as a distribution layer for Fiserv, you know, when I'm in the market trying to get a bunch of small business customers, I'm in the market competing with someone like Mercury and, you know, a few. I think it was yesterday that Mercury released a whole bunch of AI features. I think it's called command, and it has a lot of this, like Silicon Valley wrapper on top so people are able to go in. They're able to use the agent as effectively a chatbot, but with deeper LLM reasoning around financial information and then also able to use it to kind of push on buttons, right, to push functions, because the entire interface is kind of API wired into what the agent does.

And then if you zoom out, you've got folks like Ramp and Deel going even further, thinking about not just sort of AI features on top of a bank account, but rebuilding the entire financial function, you know, like rebuilding, accounting, rebuilding an AI, CFA that works with accountants in order to get things done and do reconciliation and keep the books or building out an entire agenda harness for, you know, running your company and deploying these features. And so, I guess my question to you is, how do you relate to what seems to be this extreme race in the financial AI world, and what should be the position of your clients and Fiserv relative to those other brands?

Srini Krish:
Yeah. Look, I mean, this is, you know, when fintechs first got announced, I remember I was still at JP Morgan, you know, and this is early 2000. You know, the whole idea was the fintechs are going to fundamentally disrupt, you know, the traditional banks, if you will. Right. And it's no different. I always joke about this, that since I've begun my career, there's always this talk about mainframes that are going to disappear, right. So here we are, 30 plus years later, we still run some of our most important and critical functions out of, you know, out of mainframes, right? Yes. We have, you know, sort of moved all of the other workloads outside of mainframe. But, you know, when it comes to massive scale in terms of processing speed and or sensitivity around the type of, you know, things that we need to get done. Mainframes are still around. So, the parallel that I would draw is while I think we'll continue to see places where like embedded finance, where banking services become a natural part of, you know, your day to day sort of living and, and working and going about the traditional banks are here to stay. You know, again, there's been this conversation that's been ongoing for the last 25 years about, you know, are the branches even required anymore? Right.

But yet at the same time, start an application online and be able to walk into the branch to get a few questions answered and still finish the process. How do you get, you know, a commercial loan application that used to take weeks in terms of credit underwriting and checking and so on and so forth. That can be done in a couple of days to maybe, you know, even a few hours using, you know, a generic workflow in a very sort of secure and auditable way. Right. So, I think that's kind of where this is headed, as opposed to either or where, you know, where I think, customers would like to kind of have the best of both worlds where they feel secure about banking with the traditional banking infrastructure, yet at the same time not lose sight of the fact that they want the flexibility to be able to get some of these things done in a more expeditious way.

Lex Sokolin:
And I think one of the things we've seen is also all these features do end up bubbling throughout the industry. You know, there are some folks who are spearheading them, but the time lag. I mean, it used to be five years and now it's in six months. I feel like a lot of the stuff gets pushed everywhere. One question I have for you is around Open AI, and it's from the perspective of the AI labs. So, I'm very interested, of course, in how you kind of evaluated them and why you chose them. But I'm also interested in how the AI labs, whether it's open AI or anthropic or others, how do they think about finance and financial services and what is it that they want? Like why are they going in? We've seen OpenAI partner also with plaid and building out personal financial management experiences. It's like Mint.com in 2026 and stuff happening with Perplexity, Anthropic launching all these skills across financial services and kind of deeper integrations. How do they see finance and what is it that they want?

Srini Krish:
You know, we work with both. In fact, we work with all three hyperscalers, you know, that being AWS and Azure and Google Cloud. Similarly, we work with pretty much both Anthropic and OpenAI on a day-to-day basis. We use Claude Code. We work with Clause Co-work and we work with Codex. So, I think, you know, from that standpoint, it happened to be a press release, but we actually do work with both of them, you know, from both standpoint. It is about the distribution that you reference. Right. So, you know, we tend to be in the middle as it pertains to, you know, are these by the way you work with millions of merchants in the United States and millions of merchants outside of the United States, from a from a merchant services perspective and in the acquiring side of the of our business. And just coming back to the financial services, as you rightly pointed out, for some of these smaller banks and for the credit unions, doesn't necessarily, you know, have the same wherewithal that some of the larger banks would have that can engage directly with, you know, whether it's OpenAI or Anthropic, to take advantage of some of the latest models to provide better customer service and or to be able to rationalise their process flow and be able to provide faster service to their customers.

And we end up sort of becoming the platform and the distribution channel by which they can take advantage of it. So that's, you know, to me, I think to summarize is the probably the, you know, the right sort of motivation for our own sake. Like I said, if you go back to the four bucket, you know, sort of analogy that I gave you on how we leverage it, there is enough for our own sake for us to be able to have the right partnership to do, you know, things that we do on a day in and day out basis a lot more effective in a lot more effective way that we can provide better services to our customers. So, there's plenty of angles that you know that the partnership works.

Lex Sokolin:
What about relative strength we talked about in the beginning about technology becoming cool in finance. Right. And I think part of what I have in my mind, I think about the investment banks in particular. You know, you've got the brokers and the salespeople and the capital markets people sort of in the front office. And you've got kind of it in a new Jersey data center, right? Like trying to make sure their system work. And then you fast forward 20 years and it's completely flipped, where you've got PhD technologists making billions of dollars running these hyper intelligence inference machines that are eating everything. And then they have sort of a boiler room of salespeople, you know, that they hire to go and push product. So, like that power dynamic has flipped.

And I think about the model companies, they've obviously spent an enormous amount of money to develop a very rare technological asset, which is challenged a bit with open source. And so, they're heavily going to market and trying to get into financial services and into other difficult to penetrate industries, to sort of hold some amount of ground. But in a conversation with a company like that, who do you think has the balance of power? You know, maybe today, you know, when inference is scarce and hard to replicate and these skills are scarce, and then how does that balance of power shift when you look 5 to 10 years out, when we all get used to it and it becomes commoditized?

Srini Krish:
Yeah, it's a great point. Look, I mean, I think, you know, there are these product companies that in the early stages when they're still sort of, you know, smaller to some extent in terms of, you know, color, headcount and their ability to be able to implement that product and or provide services around that product ends up creating sort of the industry around it. Right. So, and that's kind of how the services industries always are born that, you know, provides services on top of the product itself that you know, can be leveraged, right. Think, SAP, think, you know, some of the IBM products and so forth, or Microsoft for that matter. Right? So, it's no different in this space. So, you know, both OpenAI and Anthropic work with what they call as the global service integrators or GSIs, that there's quite a few of them that you could work with that understands the intricacies of how the model sort of works and functions and how you can take the best advantage of it. And, and, you know, optimize your usage in terms of token consumption and so on and so forth.

So, from that standpoint, you know, I think you may have heard this and people talk about this quite often, as opposed to the popular belief that AI would end up replacing pretty much all of the, you know, software engineers. I think the converse is potentially contemplated as well as in the universe of things that we didn't think about. Where software can be applied is going to continue to expand. You may not write a line of code, which is what historically what a software engineer is used to do. But, you know, the role might start to, you know, sort of morph into something more than just writing code in terms of designing and integrating and sort of providing the right experience and outcomes. But, you know, I think both are both are likely to happen. Right. And both could be true at the same time. So so again, you know, there's plenty of opportunities to take advantage of the underlying model. And they're both competitive in their own ways. And there are some applications that work better with one model versus the other. And like I said we use them both. We have seen great output from both of them. And, you know, there's going to be plenty of more use cases that we can take advantage of it.

Lex Sokolin:
We look deeper at agentOS and the feature set that is getting built out. I think there's a whole bunch of well-known AI or machine learning use cases around financial services, and you've talked through them, whether it's the chat interface or whether now we are wiring up those things into MCC or APIs. You know, whether it's dealing with financial functions and just putting them on loop and automating them. Like, why should I have to calculate my taxes at the end of the year if I have an agent sitting on top of the core banking system that sees every single transaction across all these businesses right to underwriting, whether in lending or other contexts. And I think it's going to take a bit of time before people are able to trust these things to do what they want the outcomes to be with, kind of the amount of discretion that they would give to an AI.

But it feels like those are fairly well-defined outcomes. What is the unknown? Unknown? Like, what are we going to see the banks becoming like? Are the banks going to continue to be just a number on a screen with a couple of buttons, or is that whole thing going to change? Where is it going? Like, how do you see this financial singularity that we seem to be drifting into.

Srini Krish:
So, look, I mean, I think with all things you want to sort of start with what you know the most and what you can. You know, what you can be most comfortable with. The probabilistic outcomes are great in sort of certain scenarios. It's never great when it comes to your money, right? It has to be deterministic when it comes to your money. Nobody likes to be guessing. You know, if my balance is, you know, 1% off or if it's, you know, 5% off, right? It has to be 100% correct every single time.

Lex Sokolin:
Absolutely. I think that's been a fascinating overview of both the banking side of the industry as well as what AI is doing to all of us. I think it's amazing that Fiserv, one of the original fintechs, has been so forward thinking in terms of integrating AI and moving quickly with these partnerships. I do think the moment is now, and all credit goes to you for pushing fast and defending the turf of your clients from what is, I think, a a pretty strong set of challengers. But it raises the bar for all customer experience when we see that in the market with that trainee. Thank you so much for joining us today.

Srini Krish:
Thank you. Lex, thank you for this opportunity. I enjoyed this conversation and you have a great podcast, so I'd like to be on it and wish you all the best. Thank you again for featuring Fiserv.