TheCUBE Research’s Dave Vellante and Scott Hebner sit down with Chris Parrish, senior data scientist at SAS, at the AWS Financial Services Symposium for a deep dive into the emerging landscape of agentic AI. In the conversation, Parrish explains why LLMs alone fall short and how agents – powered by AI, data workflows and human oversight – can drive smarter and faster business decisions in financial services.
Parrish outlines how financial institutions are navigating the hype cycle by carefully selecting use cases that justify automation through ROI and governance thresholds. He discusses the spectrum of agentic AI models, from fully autonomous systems to high-touch oversight frameworks, highlighting critical factors such as regulatory compliance, secure transactions and the evolving role of human capital in deploying these technologies.
Listeners can expect real-world insights into how cloud-native architectures and generative AI are reshaping legacy modernization, intelligent communication and customer engagement. Parrish emphasizes a measured approach to AI agent adoption, balancing innovation with risk management – laying out a clear path toward scalable, sustainable value in banking, insurance and capital markets.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
AWS Financial Services Symposium 2025. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Register For AWS Financial Services Symposium 2025
Please fill out the information below. You will recieve an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for AWS Financial Services Symposium 2025.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
AWS Financial Services Symposium 2025. If you don’t think you received an email check your
spam folder.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open this link to automatically sign into the site.
Sign in to gain access to AWS Financial Services Symposium 2025
Please sign in with LinkedIn to continue to AWS Financial Services Symposium 2025. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Chris Parrish, SAS Institute
TheCUBE Research’s Dave Vellante and Scott Hebner sit down with Chris Parrish, senior data scientist at SAS, at the AWS Financial Services Symposium for a deep dive into the emerging landscape of agentic AI. In the conversation, Parrish explains why LLMs alone fall short and how agents – powered by AI, data workflows and human oversight – can drive smarter and faster business decisions in financial services.
Parrish outlines how financial institutions are navigating the hype cycle by carefully selecting use cases that justify automation through ROI and governance thresholds. He discusses the spectrum of agentic AI models, from fully autonomous systems to high-touch oversight frameworks, highlighting critical factors such as regulatory compliance, secure transactions and the evolving role of human capital in deploying these technologies.
Listeners can expect real-world insights into how cloud-native architectures and generative AI are reshaping legacy modernization, intelligent communication and customer engagement. Parrish emphasizes a measured approach to AI agent adoption, balancing innovation with risk management – laying out a clear path toward scalable, sustainable value in banking, insurance and capital markets.
TheCUBE Research’s Dave Vellante and Scott Hebner sit down with Chris Parrish, senior data scientist at SAS, at the AWS Financial Services Symposium for a deep dive into the emerging landscape of agentic AI. In the conversation, Parrish explains why LLMs alone fall short and how agents – powered by AI, data workflows and human oversight – can drive smarter and faster business decisions in financial services.
Parrish outlines how financial institutions are navigating the hype cycle by carefully selecting use cases that justify automation through ROI a...Read more
exploreKeep Exploring
What event is being covered in theCUBE's coverage in New York City?add
What technology base underpins all of our products today and has been evolving to be accessible through different interfaces and data sources?add
What companies have transitioned to AWS and are leveraging a platform on AWS?add
What is the company's strategy for having their customers take advantage of scaling in the cloud with their software?add
>> Hi, everybody, welcome to New York City. This is theCUBE's coverage of the AWS Financial Services Symposium. My name is Dave Vellante. I'm here with Scott Hebner, who is co-hosting today. Yeah, we just came off the keynote, Scott Mullen's keynote. He described that they first had the symposium in 2015 at the Roosevelt Hotel, which I think is closed, a small pop-up, and I think they had 35 people. There must have been at least six, 700 people, maybe five, or 600 people in the keynote, and a bunch of people milling around in the audience. And so we're super excited. I mean, it's all about how financial services is adopting not only the cloud, and the cloud services, and the fundamental cloud services, but also numerous other agentic agent building platforms leveraging bedrock. So, a lot of discussion around that innovation. We're pleased to have Chris Parrish here as the senior data scientist at SAS Institute coming off a big showdown in Orlando.
Chris Parrish
>> Yeah, thank you.
Dave Vellante
>> Good to see you. Thanks so much for coming on the program.
Chris Parrish
>> I appreciate it. Thanks for bringing it.
Dave Vellante
>> So, we know SAS, I mean, I was commenting that's what we used in college, and we were all trained on it, and still to this day have fond memories. And you guys have evolved dramatically. I've been to some of your events, and you heavily invested in technology, and AI. Where do you fit in this whole AI spectrum?
Chris Parrish
>> Yeah, so I mean, we're definitely evolving, and the platform is moving with the customers to the cloud. And so that's the SAS asset that you've probably have been familiar with is one primarily of a language that's built on top of functions, and algorithms. And that underpins all of our products today. And so what we've been working on for the past 10 years, and perhaps even longer, is to be able to take the technology to the cloud with our customers moving next to their data. And so the types of applications that we're building today are still using that foundational SAS as the technology base. We're just making that more accessible through different interfaces as well as connections with different data sources, as well as being able to enhance our current product suite to make those purpose-built solutions for both financial services customers, and other customers in our portfolio.
Dave Vellante
>> So, simplifying the experience, but also supercharging it?
Chris Parrish
>> Yeah, it's supercharging, and bringing in more users. I mean, that's really the goal. The SAS language itself is really just a very small component of the bigger platform. I mean, we do have a very large install base of SAS users, but clearly there are new technologies out, and we have to connect with them. And so that's what we're doing to be able to have users really using any other tool to plug into our ecosystem to eventually get to the business decisions that they want to get it. And that's what we've been doing best for all these years.
Dave Vellante
>> So, when LLMs first hit the scene, it was pretty obvious that people were excited, but then they try to apply them, and then they say, "Oh, wait a minute." Especially in financial services. And I know, Scott, you were just down at the conference, and I'm sure you've heard a lot about this. So, I want to get into, maybe you could summarize the conference a little bit, what you learned. And then I want to understand from SAS, why do LLMs fall short, and what are you guys doing to build that robustness, and that compliance, and that non-hallucinative capability? But Scott, why don't you kick it off with your quick summary.
Scott Hebner
>> Yeah, I thought you guys did a really nice job at the conference. I mean, I think you took your customers on a tour from 50 years of history as a powerhouse, and data analytics. Then you got into the conversation about LLMs, and then into agentic AI, and agents, digital twins, and what if simulations right up through quantum AI. And you pulled that all together, but you framed it, and I remember the graphic, it was a rainbow with the unicorns about LLMs, right? And making the point that they're insufficient alone because your overriding theme was decision intelligence. So
Chris Parrish
>> Yeah, that is right. And the technology evolves, and as a software company, we have to evolve with it. And so we do recognize that there are different hype cycles as technology evolves, and there's life cycles to that as well. And so I think that the unicorn type of graphic was really there to say there isn't any easy buttons in this, and there's not something that is going to magically change your business. It's really going to have to be an integration with a variety of different services, or different types of controls that you're putting in place. And so that's the picture we're painting is these are great tools. They're incredibly innovative, they will continue to get better, but we see them as a piece of a bigger puzzle. And that's where sort of the agentic AI kind of comes into play in that you're... The way I think about it, the way we think about agentic, again, myself, personally, is it's really just an application of AI. When you're automating processes, and you're trying to find those use cases that are initially low risk, high return, and how can you extract costs out of those processes? And to do that, sending something to an LLM, and getting something back is probably not going to do that for you. You're going to have to orchestrate that with rules, with maybe traditional models, with data queries, with workflows, all culminating into one sort of super decision that gets you that answer that's governed, that has the appropriate level of automation with human in the loop, and that has that orchestration tool that helps you put all that together.
Dave Vellante
>> Scott Mullins, in his keynote, he referenced Arthur C. Clark, who's an author, and a futurist. I've heard this recently in this whole AI wave. He's got three laws. His third law is, quote, "Any sufficiently advanced technology is indistinguishable from magic." And when you think about agents, it sounds like magic, although you just mentioned a number of things that aren't magic, it's sort of core technology that enables you to create what seems like magic. I wonder if you could double click on that.
Chris Parrish
>> So, what's likely going to happen, I'm not prognosticating, but I think what's going to happen is that as companies sort of assess those use cases, they're going to want to have some type of expectation of what those use cases are going to return for them. And so to be able to do that, you have to show that you're putting some governance around that, and putting some guardrails, and some failsafe. So, maybe if you look at the world we live in today, there's full automation, or near full automation maybe in driving, and other features, but there's still some fail safes, and there's ways to sort of stop the system if it needs to stop. And that's where those, I would say, more bespoke type of processes will come into play. And you can even see this evolving into agents calling agents, calling agents, calling agents calling... And that's where our software is built to embed not just a one, and done decision, but multiple agents within a super decision. And so as they evolve, and become more, I think as customers become more, users become more comfortable with that concept, and they have the right governance, then they're going to begin to say, "Well, I have this agent that does this, and this agent that does this", and how do they all talk to each other? So, it's an orchestration of it.
Scott Hebner
>> You guys really hit the trust factor really nicely last week. Without trust, it's the currency of innovation, right? No trust, no ROI, especially if you're making decisions. And I thought you did two things really well one is the introduction of Epic Games into the digital twin simulations because you can visualize it, you can see it, right? That's going to enhance people's trust. I also felt you told a good story about the verticalization of the models, and the agents you guys are going to build. Very domain specific because if you're going to try to help emulate how humans think, and make decisions, you have to understand those domains. So, covering all that around the core via platform, and across really well, I thought.
Chris Parrish
>> So, I mean that's been the core... The agentic piece of it, we've been doing sort of agentic analysis, or agentic decisioning for many, many years, and that's built into our core technology. So, when you buy our platform, you can get that sort of embedded with it. So, it's not a futuristic kind of technology we have it today. What's different is what can I pull into that decision? What kind of new technology, what kind of, how can I leverage LLMs, or some other generative AI technology to enhance that process so that I'm automating it even further? So, agentic is really going to be an automation tool, but it's going to be an automation leveraging even more sophisticated technology so that you can extract even greater cost.
Dave Vellante
>> So, what does AI bring to that equation? Is it are you able to interact with the system through natural language? Is it also taking probabilistic AI, and somehow applying it to maybe tighten up some of the false positive? How does gen AI fit into that? Because you're saying you've been doing agentic workflows for a while, so with, sorry, traditional machine learning .
Chris Parrish
>> Right, or statistics, or even just rules, right?
Dave Vellante
>> Okay. So, what does gen AI bring?
Chris Parrish
>> Well, it brings it to it, so, it has where you need to do maybe some type of large summarization. So, say there's a lot of data coming out, unstructured data that the companies have that it's almost impossible to kind of go in, and get some type of summary from it without using some type of large language. So, the large language will bring that sort of technology into the process. So, you can begin to look at what types of prompts I want to then assess, or put into that LLM. And so that means that you can now begin to manage those prompts, or you can begin to manage, maybe you want to select a number of large language models, maybe you want to compare them, maybe you want to do AB testing on them. Which one gives me the better result? But what it brings in is the ability to take a lot of that unstructured data, and pull it into those, I would say, traditional type of agentic decisioning roles primarily has been primarily been rules based, or advancing to statistical, and machine learning models.
Scott Hebner
>> There was an article in New York Times a couple of weeks ago on how the LLMs are getting more, and more called reasoning chain of thought, things of that nature. And actually the hallucination rates are going up, and I think it's because I think it was the rainbow going back to that unicorn statement that generative AIs, and LLMs are not really designed to help you make decisions. They're correlation based. And again, I think what you guys are doing is you're building upon that as a core foundation, and engine, adding the decision intelligence on top of it with the trust factor, and the governance, and the guardrails, and with your history as you know around analytics, and the partnerships you have with AWS, and others.
Chris Parrish
>> Yeah. So, the idea is as you look at these use cases, and say do I even...? The first question is, do I even want to use this type of technology in my process? And so that's the sort of first gating question. Then the other question is what kind of automation do I want in the process, and how do I govern that? And governance being like, what are your tolerance? What's your pain thresholds? Where are your fail safes in that so that you're not... A lot of things that we're seeing now today with our customers talking to customers across the world is that they have lots of ideas, and they have lots of interest, but there's not really a way to sort of tactically put all that together. So, we want to come in, and say, "You're looking at this technology as fantastic technology." How do you tactically put that in place so that you are getting the kind of results that you want? What's a good enough result? Is 60% good enough? Probably not. Is 80% good enough? I don't know. So, those are things that the companies have to determine internally to say, what is my pain threshold, so to speak, for this particular project that I'm using agentic AI, and how can I convince the people I need to convince that it's sufficient to be able to do that? So, we're working with customers all over the world really to try to understand what the use cases are. And so I would say we're in the early stages, and that's to be expected.
Dave Vellante
>> So, that involves what you describe some trial, and error. Yeah, hey, check it out. Well, 60% is not good enough. We know that. Okay, let's try 75%. What kind of feedback are you getting from your users? You start to scale it a little bit, and maybe we need to tighten that up a little bit. So, the cloud makes that easy. Experimentation. What is your relationship with AWS? And I want to come back to sort of, you're suggesting that we're early days, but I want to dig into that a little bit.
Chris Parrish
>> Yeah. So, broadly speaking, our relationship with AWS is a strong partnership. We have transitioned a lot of our clients to, so traditional SAS for those not familiar, is typically installed in data centers, and servers, and it's run through their administrators. And obviously some of that that's changing. And so we're working with a lot of our customers, some of our banking customers like Jyske Bank in Denmark, and we have some insurance customers, SBI in India, and we have Toyota Financial Services in Italy that have all sort of moved to AWS, and are leveraging our platform on AWS. And so that's really our goal is for those customers that are ready, and are wanting to move to the cloud, and AWS is their partner, we're helping them make those migrations. We also have a variety of, and what makes us different probably than a lot of other software companies is that we have all these purpose-built solutions as well. So, we have risk management solutions, we have marketing solutions, we have fraud solutions, and those are purpose-built for those particular domains. And those are all can run on AWS as well.
Dave Vellante
>> So, you mentioned it's early days, but we're now a few years in to this AI awakening. Let's say it's a nine inning game of, we have third of the way through. It's clearly not the top of the first anymore. I mean, we've gone through a lot of POCs, been a lot of experimentation. Sometimes the narrative is, hey, the people are stuck in POC purgatory. You listen to the AWS keynotes, like there's real use cases actually going on to talk to some of the people here. So, the truth is kind of somewhere in the middle. How do you see it?
Chris Parrish
>> Yeah, so I think there is large would say, and this is anecdotal, right? So, we're just talking to people, but I would say that most of, and this is probably focusing on the size of the organization as well, but I would say the larger financial service organizations have budget, and they want to be able to access this technology. And so what they're trying to do is find those internal use cases that are going to help them identify cost savings, primarily. And so there's probably going to be those situations where you can extract some type of proxy FTE reduction of some sort. And so those probably aren't going to be use cases where you're making decisions that are going to impact the customer directly. Or if you do, then there's obviously some more human sort of in the loop in those situations. So, what we're seeing with our larger customers is that they're building out some systems to be able to extract cost savings from certain processes, whether it be collections, or customer support. Those things internally that I think are probably less regulated, and they can have a little bit more leeway, and kind of feel their way through the system, and determine what type of risk appetite they have for adopting these types of systems. And I'd say for some of our smaller customers, it's really just trying to automate some of their daily tasks. And we're building that into our platform as well. So, within the platform we have, we're building up Co-pilots, you can say, build me a dashboard that shows me the sales trajectory, and it builds out the dashboard.
Dave Vellante
>> So, thinking about SAS generally, and the financial services opportunities specifically, where do you want to be a year from now? What do you want to be able to say a year from now that you can't say today?
Chris Parrish
>> Well, we'd like to be able to have a lot of our customers take advantage of the scaling in the cloud with our software. I think that we've done a great job in our R&D side, and we have incredible technology. You guys have talked to the folks that are in the team, and we have incredible applications, and ability to take their work to the next level. And so we're working with customers closely through our partners like AWS, and some of our other consulting partners to get that message out to our customers to say that this isn't just about a SAS language. This is about accelerating what you're doing, and making businesses trusted business decisions. And so that's our path forward is to be able to send that message, and to have our customers migrate to a platform that's modern, and that fits their needs.
Dave Vellante
>> We're seeing a breakthrough in financial services. Things are getting faster, but at the same time they can't break especially in this industry. Chris, thanks so much.
Chris Parrish
>> Thank you very much. Appreciate it.
Dave Vellante
>> Really appreciate it. Okay. Thank you for watching. Keep it right there. This is theCUBE at AWS Financial Services Symposium from New York City. We'll be right back after this short break.
Dave Vellante
Scott Hebner