Exploring agentic tools with Ravi Nemalikanti (at AWS Re:Invent)

For the last episode of 2025, get to know Abrigo's CTO, Ravi Nemalikanti, as he talks about his AI philosophy at Amazon's AWS Re:Invent conference. Listen in to learn about the metrics Abrigo considers when making decisions about machine learning in its solutions, ensuring that those decisions support community banks and credit unions. 

About the guest: 

Ravi Nemalikanti is Abrigo’s Chief Product and Technology Officer and is responsible for leading technology strategy and determining product and development priorities to drive innovation and increase the company’s competitive advantage. Ravi is the Winner of the 2024 Haas Technology Leadership Awardee for North America by Carlyle, an award given to celebrate an exceptional technology leader. Before joining Abrigo in 2022, Ravi was the CTO of Digital Banking at NCR Corp., where he led the organization’s digital-first banking technology roadmap. Earlier, he held leadership roles in Tax and accounting, Global Trade, and Risk Management during 14 years at Thomson Reuters. Ravi holds a bachelor’s degree in engineering from Andhra University in Andhra Pradesh, India, and an MBA from the University of Chicago’s Booth School of Business.

Helpful links: 

AI Hub - Abrigo

Webinar: AI strategy for banking: Unlock the most value - Abrigo

Transcript

SPEAKER_00 0:00

Today on the AWS for Software Companies podcast, we are welcoming Ravi Nimalikanti, Chief Product Technology Officer for Abrego. Welcome to the show, Ravi. Thank you. Great to be here. Great to have you. Now, for folks who may not be as familiar with Abrego, can you spend a moment talking a little bit about the organization, the customers, and the problems that Abrego helps them solve?

SPEAKER_01 0:23

Obrego serves about 2,400 banks and credit unions in the US. So that's about one in four financial institutions in the U.S. So we have quite a bit of presence in the states. And our mission is quite simple. We want to be and are the technology arm and the innovator that these community institutions rely on. So if you step back and think about the mega large financial institutions like the Chases, the Bank of America's, they're spending billions of dollars on the technology budgets. Community banks don't have that kind of a budget. So what we do for them is to truly bring the power of innovation as a technology partner, not just another vendor that's serving a point solution. So that's that's what we do for these institutions. To the specific products and solutions that we offer in our space, um, again, it's very simple. We help them manage their risk and help them grow responsibly and safely. So what that means is we help them with their anti-money laundering products and services. Um we help them with detecting fraud, we help them with asset liability management. So all of that on the risk management side. And on the growth side, we help them originate loans in a digital first kind of way. So from commercial lending to small business lending to consumer lending, right? So we are the one partner that provides them all of these different platforms that delivers all of these products and services, as well as an intricate look at their data sets. Again, as uh as these community banks that have been around for a long time, there's a lot of legacy systems in play, right? And the data is very siloed. So we also help them with um with visibility into their data using our Connect platform as a data analytics and visualization platform as well. So again, our mission is to be the technology uh enabler for these banks and credit unions, and that's what we do.

SPEAKER_00 2:32

Okay, so let's say I own a couple of credit unions in Topeka, Kansas. Yeah. You know, small operations, been there a while, very uh welcomed and trusted by the community. We call up Abrego. There's the Connect platform that will get us moving from paper to digital. Um, there's some consultancy that would help mature our data storage.

SPEAKER_01 2:54

Yeah. Is that essentially the Yeah, yeah, that's that's uh definitely one way to think about it, right? So you would you would either call us and say, Hey, we see customers or members not coming into the branches anymore, and they want more digital first experiences, whether it's a small business that they may have worked for the last 20 years, but now they need a bridge loan and they're starting to go to the upstarts that are providing these services and can also go from an application to cash within within a few hours, right? Whereas their existing process may not be uh helping them do that. So um whereas this business that you're serving needs cash now. Maybe they need to buy inventory to to serve their customers and so on. So those are the transformational changes that are happening, whether it's Topeka, Kansas, or you know, mm San Francisco. This is the same challenges. The demographics are changing, the expectations are changing, people are now used to Uber type experiences. Like it's that instant gratification. I need an answer now. Will I get the loan now? If I if I'm not eligible for the loan, uh what do I need to do to make sure that I am eligible for the loan? Right. So those are the types of things that they would reach out to us. But then again, they're they'll have their own credit office, they have their their policies. What we are helping them do is put all of that in a digital first web experiences, mobile experiences as part of our loan origination uh suite. Connect platform, on the other hand, is to really bring together data to say, how are my, for example, right, if you are a chief lending officer of these two credit unions and you want to see, hey, which branch has a longer, deeper pipeline for loans? And why is that better performing than the other bank? Well, you go to Connect to get those answers.

SPEAKER_00 4:53

Oh, that's really helpful. Thanks for explaining that. So how is AI transforming your product roadmap these days?

SPEAKER_01 5:01

Yeah, I mean, it's a fascinating time to be in technology. Um and I've been in technology now for about 24, 25 years, and it always seems that n this is the pinnacle of technology. And uh, we've seen a few of those waves, right? If you look back in the last 20 plus years, iPhone and how iPhone changed the technology curve and adoption, especially in financial services, right? Now everybody walks with a with a mini branch in your pocket. Now you can apply for loans, you can check and check your balances, you can now send money to your peer, right? You sell, you name it, right? So you were walking uh with a with a mini digital branch in your pockets. So we've seen that wave, and then the next wave was equally transformational from a cloud perspective, what AWS, what Azure, Google Cloud, all these hyperscalers have done in really bringing that compute to the next level, right? Bringing that storage and the services and how quickly you can go to market with any new ideas to that lightning speed that we have seen. And then in the last 18, 24 months, we have seen AI just take it now 10x, right? With respect to how we have to one reimagine our experiences. I mean, chat GPT has just completely changed how we think about conversational experiences. I mean, two years ago, you'll have to really look hard for somebody that loves talking to a chatbot.

SPEAKER_00 6:36

Right.

SPEAKER_01 6:37

Uh now I find myself going to Chat GPT or no pick your favorite conversational tool, right? They come a long way with these foundational LLMs. And that is truly changing how customers again, how we've gotten to that instant gratification with Uber type experiences, everything at our fingertips. Now we want you want to search your knowledge very quickly. If I have you know multiple policies, you know, take uh AML policy, a credit policy as an example, right? And there's 15, 20 different big books of policies that these financial institutions have, how do I get to the answers quickly? How can I search that knowledge quickly? That's a that was a huge barrier, both from a very simple new employee on voting perspective. How quickly can this teller at a branch come up to speed with my policies? And by the way, Tellers is one of the most uh uh at-rating job families within the within the bank and banks and the credit union sector. So, how do you actually keep up with that? How do you make sure that you can bring employees up to speed quickly? So, those are the types of very simple but highly effective use cases that AI is completely changing. Now you can actually give them an iPad or some tablet device that has conversational experience into all of the policies. All you have to do is, for example, if you are a new teller at this branch in Topeka and somebody walked in with a temporary ID and said, Hey, I want to cash this check. What is the right thing to do? Can I cash a check? Can I give them the the cash for this check with a temporary ID? Like, what's the policy? In the past, you have to go find somebody and ask the person, you're a new employee, you know. Right. Or you do the wrong thing and you actually help them, you know, say a fraudulent check. So that's the that's the difference where true value proposition, excuse me, from a from an AI perspective, that changes how you onboard employees. So it's a very simple but effective example on uh in where AI is truly already changing the landscape.

SPEAKER_00 8:51

But but as the chief product technology officer, yeah. I'm sure your roadmap has been disrupted many a times on this. How do you balance that?

SPEAKER_01 9:00

Yeah, so it's a it's a great question that we wrestle with because it's never a done. Do you oh now you fix the roadmap because of this new technology?

SPEAKER_00 9:09

Uh it's a it's a journey. Or something's just come up and it's like, oh my gosh, that's gonna change our three-month outlook.

SPEAKER_01 9:15

Yeah, I I'll I'll tell you what really changed for us as a company at Obrego, and that actually is um is from a uh from a disruption perspective, it's the true innovators in dilemma, right? Because we have had these products for many, many years. We continue to innovate, and those are all incremental innovations, enhancements for you know thousands of users at these banks and credit unions that know how to work with the platform within their workflows. Now, what's started to happen is to to your question around what truly changed in the last 12 to 18 months is the cost of POCs and prototypes has literally dropped down to zero. What used to take, you know, a few people in a garage somewhere to come up with an idea and build the idea, it would have taken like six months to actually get to a product that they can show. Now it's a weekend and a couple people. Yeah. So what we are already seeing is even though we are one of the the major incumbents and we have the track record, we are starting into conversation, we are walking into conversations where prospects would say, ooh, that looks interesting, and that's actually a very, very small startup company that is very quickly built this prototype that seems like it works well. So there's a lot to be you know unraveled there because though some of the things that don't uh go along with that, you know, the five vice that you know you need to ask when somebody does it is what's the governance? How is the quality of these applications? How do you maintain it? How do you operate it? So there's all this that last mile in delivering software has still holds up, but you start to get into prospect conversations where your demos, your products don't look as good because somebody's just built a very custom demo for the very specific use case that this bank is looking for, right? So that's what really changed for us. But how that manifests for us is as you step back and look at, you know, I'll give you a specific loan origination example. So if you're a you know$10 billion uh asset center management bank, you know, you may have anywhere from five to twelve people that touch a loan application, from a relationship manager to a credit analyst to an underwriter to a loan reviewer, closer. There will be around five to ten people that touch um that particular application. What are the activities that are happening within those personas and how many of those activities are truly disrupted by AI? Is where we are spending the most time because what we can do is because we have all of this data, not only about the the information that's flowing through the system, but how our customers are using our existing platforms and what are those friction points. And and that's where we've leaned heavily into to remove those friction points. For example, what is the activity that's happening within the system? What is the activity that's happening outside the system? What do I need to do to bring the activity that's outside the system into the system? Or at least controllable. Control it. Exactly. So those are the types of activities where we had to go back and change the roadmap and say, look, those incremental enhancements are good, but we need to really reimagine the whole process. Does every single person, these no five to ten people, do they have to touch the uh application? Or is there a better way? Yeah, maybe we reimagine exactly. So that's what's changing our roadmap. So we put out multiple products um that tackle exactly that in the last 12 months. Okay. We've uh we put out a product that helps a BSA analyst, Banking Uh Secrecy Act, right? Um, that's um that looks at transactions that come in on a daily basis and see which ones have alerted their policies against their policies. And they have to go research the transaction, research the account, and then write a summary for the alert. What we said is, well, it takes about, depending on the complexity of the case, it could take anywhere from 15 minutes to a couple of hours. Can we give them a better start? If it takes 15 minutes for the manually, can I give them a summary like within 30 seconds?

SPEAKER_00 13:44

But what what are the measurements that you typically use? Is it speed of transaction? What are some of the like your customers, what are they kind of looking at? What's on their dashboard?

SPEAKER_01 13:53

Yeah, so that's a great uh question because we um as a as a company, just like you know, AWS is we are very, very customer obsessed, right? Uh we always start with what is the true value proposition? Um and even in our sales motion, one of the things that we talk about and share is the ROI. So if you're spending X dollars with us, what is the ROI for you? What is the efficiency? What are the processes that maybe you need to get rid of, rethink, reimagine, all of that? Coming back to some of the specific metrics that we look at, a case alert and how long does it take to write a narrative for a case alert? It takes you 15 minutes, can I give it to you for free? Like zero time spent on it. And you go change it. Because if you think about just step back and think about some of these metrics, right? The core metric is always for us is can we save a minute off of the grunt work that an employee at a bank or a credit union does? Because our core segment is the community banking segment, and in the community banking segment specifically, relationships are the most important, right? They know what Joe's coffee shop is doing, how they are doing. They want to spend more time on that. They are the ones that are helping local economies thrive, right, as the banking institutions. So if we can take a minute minute here, a minute there, and help them just redeploy that minute in understanding their customers or their members better, it's mission fulfilled for us. Right. So every single aspect that we do is truly, even going back to our origination of our company, right? The inception of our company, it's about automation. Back then it was workflows, stitching together workflows, putting together low-code, no-code automation platforms on top. That's what we have done. Again, in spirit of decreasing the amount of time they have to spend on the grunt work so they can go build the relationships, grow the bank.

SPEAKER_00 15:53

Good. Okay. So that makes sense. So speed of transactions are absolutely key to that. Um, I'm sure that marries together with uh satisfaction and some other metrics that are on the dashboard. How about the large variance in potential customers of your customers? For example, uh, members of various credit unions, they may have been there 20, 30, 40 years. They may be of the age where this technology can be quite baffling to them. How does a Brigo uh compensate for that? Or or do you get involved in that user experience at all?

SPEAKER_01 16:30

Uh yes, we do. Uh to the extent where you know someone's applying for a loan. What does that borrower experience look like? And how much of that friction can we remove so there aren't a whole lot of uh you know dropped cuts, so to speak, in a retail sense, right? So you start applying for a loan, but the process is so intricate that you abandon the application, right? We want to decrease that. So that's an important metric that we track and our customers track as well. To the question that you're posing, which is how is the change in demographic affecting the financial institutions? Our goal is to keep it very simple. So irrespective of whether you're a Gen Y or millennial or no baby boomer, the experience is very, very simple. It's very um intuitive and most minimally required information that that we need to have to make a decision on a loan. And that's the only area where we we have the borrower experience, right? Otherwise, it's our our products are all internal faced. So for the banks, so it's the BSA analyst, it's the credit analyst, it's it's the portfolio um risk managers, loan review officers. So they use our products the most, except for when you're when you're applying for a loan, we manage the borrower experience, which keep it extremely simple, ask for the most minimally required information in order to get to that decision. Again, because these are these are such established relationships in the community banking sector, they usually know, even before they started applying for the loan, they probably had this conversation already. Right. Right with the with the relationship manager. Right.

SPEAKER_00 18:22

Yeah. So when you start to look ahead next 12, 18 months, how do you see this part of the business evolving, especially as you start to see agencai AI in the forefront? How would you imagine the next year or so will be on the industry?

SPEAKER_01 18:39

We are very excited about uh the possibilities of agentic AI and some of the uh announcements that we've heard at AWS reInvents, uh reinforced uh our our excitement and emphasis that we've been placing on agentic AI internally uh as a company. Uh the reason being there's a today if you have to build an agent, even the pace at which this has been moving, this whole space has been moving, and we started building uh agentic flows a few months ago. So we started we were very early users of strands, AWS trands, early users of bedrock, and we also started using agent core very recently, right? Uh some of the things that we've learned along the way in a very painful fashion is as you're trying to stay ahead or keep up with where the market is and the pace at which this is this is going, there's a lot of new non-functional things that you have to start thinking about, right? Like how how does a model drift or quality of uh a response, how do you make sure that's in a range, right? This is not typical quality analysis process or assurance process that we've uh we've put in place 15 years ago, right? You have Think about this in a in a new in a new way. What type of guardrails are required? Because we we put out a a new product called Ascobrigo, which is a search, conversational search into policy tools and manuals that uh these financial institutions have. And we went through that journey where, oh, we have to think about all these other guardrails that maybe we didn't need to think about in the past as we as we started deploying uh different agentic tools. One of the tools was okay, I don't have the answer to the question that was posed in the knowledge base. Can I go to the browser or can I go to the internet and search for an answer? What type of a consent is required? Because as soon as you go out of the boundaries, you might even you might be hallucinating, you might go to a wrong source. Right. So how do you control that, right? So you have to start thinking about all of these guardrails that just with some of the announcements that are coming out with agent core, with the policy, which is much more deterministic in nature, right? You can set very good ground rules in uh that you cannot break, or uh agent core evaluations from a quality perspective, all of these this is music to our ears because this accelerates our development lifecycle in how in how we think about um agentic AI. And one of the things that we've been focused on is as we step back and reimagine some of these more deterministic workflows that we've had, how deterministic do they need to be in the future? What are the indeterministic aspects of it? Can they be replaced by agents? And if so, what does that orchestration layer look like? And we felt like we got a lot of those answers um in the last couple of months. So we are really excited about how agent TKI can truly push us and the industry to reimagine some of these uh experiences.

SPEAKER_00 21:55

Fantastic. Absolutely so are we, Ravi. Uh again, I want to thank you for uh for joining us here on the podcast. Where can people learn more about Abrego?

SPEAKER_01 22:05

Uh we most certainly uh put out a lot of content on LinkedIn. We also hold a lot of webinars on the topic. Again, to the trusted advisor of these community banks, right? We like to educate ourselves and our customers in the process. So uh webinars, LinkedIn. I personally post a lot of articles on what I am learning from these conferences and uh um and these conversations, frankly. Um So that's that's where you can find us, and of course at our website, abrego.com.

SPEAKER_00 22:37

Thank you very much, Ravi. And again, best of luck. Look forward to speaking with you in twenty twenty seven and uh talking about what the future's like.

SPEAKER_01 22:45

Well, it was uh it was a pleasure. Thank you for having me.