We just sent you a verification email. Please verify your account to gain access to
AWS re:Invent 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 the link to automatically sign into the site.
Register for AWS re:Invent 2025
Please fill out the information below. You will receive 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 re:Invent 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 re:Invent 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 the link to automatically sign into the site.
Sign in to gain access to AWS re:Invent 2025
Please sign in with LinkedIn to continue to AWS re:Invent 2025. Signing in with LinkedIn ensures a professional environment.
In this segment from AWS re:Invent, Anand Raman and Jeff Veis from Impetus Technologies join theCUBE’s Dave Vellante to discuss the critical "data plumbing" required to support the rapid shift from generative to agentic AI. The conversation highlights the reality that there is "no free lunch" regarding AI adoption; enterprises must first modernize legacy systems and optimize cloud infrastructure to handle the scale of automated decision-making. Raman and Veis unpack the company’s "3 Es" framework – Engineering, Economics and Experience – explaining how CIOs c...Read more
exploreKeep Exploring
What is the current state of data's role in business, particularly in relation to advancements like generative AI and agentic AI, and how does this relate to the evolution of a specific company?add
What are the challenges enterprises face in modernizing their systems and integrating older technologies with newer advancements like AI?add
What are the critical aspects that enterprises consider as engineering evolves in the context of increasing costs associated with cloud systems?add
What is the significance of Prism in relation to observability across data platforms and hyperscalers?add
>> Hi everybody. Welcome back to Las Vegas. We're here tucked away in The Venetian in suite 3708. You'll come by and see us, third floor of the Venetian. My name is Dave Vellante and you're watching theCUBE's live coverage of re:Invent 2025. 13 years of re:Invent. We've been here, theCUBE's been here 12 years in a row now. The one exception was COVID. But we covered re:Invent there remotely as well. We're super excited to have a couple of CUBE alums back. Anand Raman is the CRO of Impetus Technologies and he's joined by Jeff Veis, longtime CUBE alum. He's the CMO of Impetus. Guys, we're going to talk about solving complex data problems, which is what your superpower is. That is the number one challenge, I think, today in AI, but welcome. Thanks for coming in.
Anand Raman
>> Thank you, Dave. Really appreciate it.
Dave Vellante
>> It's great. AWS, amazing. This event is incredible, as you well know. AWS puts out all the tooling and even the frameworks, when you think about SageMaker and Bedrock, for developers. But there's still a lot of work around data that has to get done. Gen AI has completely changed the focus. We always cared about data. You and I have been talking about data for decades, but now it's a whole new game, isn't it? Well, I think fill us in.
Jeff Veis
>> I think so. There's no free lunch. So the hard work to modernize, to be able to deliver that value, to be able to take advantage of what gen AI and now agentic AI, right? Gen AI, that only lasted for about three months, it seemed, right? We're onto the next kind of term and focus. But that just builds on everything that we've kind of focused on. And we've seen our business just explode because data is at that center. That hasn't changed.
Dave Vellante
>> So for people who don't know Impetus, explain your business. When you meet somebody in the elevator or the CEO of the corner suite, how do you explain the company?
Anand Raman
>> So we are a very niche organization and providing data and cloud engineering with now gen AI to enterprises. So we help them build next gen applications with a very, very solid data platform, data plumbing system behind that can scale across millions of users, hundreds and thousands of petabytes of data.
Dave Vellante
>> You know what's interesting is, I think it was Matt Garman maybe in his keynote, might have been somebody else, basically making the case, AWS has a challenge. They've got this, now, I hate to use the word legacy, but they've got this legacy install base. They're like the new legacy with millions of customers doing virtualization essentially in the cloud, compute storage, networking. And now they've got to bring in AI, gen AI, and they can't just bolt it on. Everything has to ... The whole stack has to be refactored. But they made the case that you've still got to modernize before you can take advantage of gen AI. So what does that mean for your customers in terms of how they're responding to the new data challenges?
Anand Raman
>> Actually, a fantastic question, and thank you for bringing it up. Modernization has been our cornerstone of how we are helping enterprises. And modernization is coming in multiple aspects or respects, if you would, for an enterprise. We still talk to biggest of financial companies who are still using mainframes and Oracle systems and others who want to get out of it. We've people who have implemented, like you said, they've gone onto AWS, but the systems are not optimized. And now with AI, gen AI, and agentic coming in, it's becoming extremely difficult for them to stitch all these three different systems. One is probably 40 years old, the other one is probably 10 years old, and now gen AI coming in, which is two years old. How do you stitch all these systems together? So our modernization has been, "Hey, eliminate the past and get into the new one." So when we were here at the last session, I spoke to John about, our concept has been, how do you modernize an organization? It's through engineering, it's through economics, and it's through experience. That's the E that Impetus stands for, engineering, economics, and experience. You have to have good engineering, it has to provide the economics for the enterprise, and the customer experience has to be fabulous. So on that line is what we've been working through over the past few years. And while engineering is evolving and maturing on a daily basis with everything happening, economics and experience are the two critical aspects that every enterprise is wanting to do. And economics, with everything that's happening on AWS, or any other cloud for that matter, while they have put in systems, either legacy bolted on, lift and shift, or new systems, the cost of all these systems are all going up. And from an economics perspective, a CIO has to make a justification on how do I control that cost? And that cost is not going to be contained. That cost is going to go up, but the cost goes up with the new applications and the AI and gen AI. And with everything that's happening, they want to make sure that the cost of building agentic systems is also controlled. So what we are doing now, from a CIO experience level perspective, is saying, how do we bring economics to this, to this entire ecosystem? And as Jeff was talking about, what we've built out is a system called Prism, which really provides the economics of the entire cloud ecosystem.
Dave Vellante
>> Okay. So let's back up a little bit because I want to set the tone.
Anand Raman
>> Sure.
Dave Vellante
>> When you think about, because we all remember RPA, the data set was we're going to watch a human do clicks and then we're going to repeat that and take these repetitive tasks away. It was kind of the early days of ... Well, I mean, there was always that automation, but that was sort of the next generation automation. And it drove some value. A lot of the similar dynamics, people were afraid it was going to take their jobs away. And it was pretty deterministic. You could trust those software robots. And then you saw agentic, it was this request-respond. I remember George Gilbert and I wrote a piece a couple years ago now that the request responder era is over. It's now all about agentic and systems of agency. And so systems start acting, not just responding to queries. That's a whole different ballgame, Jeff. And presumably Prism fits into that evolution.
Jeff Veis
>> Yeah. No, to put a cherry on what you just said, we've spent the past 20 years automating tasks, okay? And that brought value, it allowed scale, it had benefits. But now it's a different mindset. It's about automating decisions, and decisions in terms of the agentic agents that are sometimes working together, multiple agents. And it's understanding not just what happened. And that's the big aha. Most observability or cost management solutions, rear view mirrors. They would tell you what occurred, you'd get that monthly report and you'd go, "Oh, department X spent 30% more. That's a red flag. We have to address that." And that gave you some controls or at least the perception of control. But now it's a different ballgame. Now it's being able to understand where you're tracking, proactively identify anomalies. Very different than just detecting anomalies after they occurred. And then, and this is what's going to separate the adults from the children, predictively or proactively taking that action. Because you hear a lot of metaphors now, but they're saying agents are like giving a credit card to a teenager and going away for the weekend and hoping a good outcome comes. And I'm not accusing all IT staffs of being teenagers, but you have these agents that are willing to scale, willing to consume and willing to act, and you need to be able to take advantage of what they can do, but you need to be able to have some way to both unify, optimize, and predict what's going to occur.
Dave Vellante
>> So the way you describe it really aligns with our research. We're good at what happened, okay at why it happened, really not good at what's going to happen and what to do next, kind of the next best action. And when you talk to companies that are deploying that, Amazon itself, companies like Dell, companies like JPMC, they have to go through a really onerous data exercise to get it right. And I see some of the predictions, you see Gartner, 40% of enterprises are going to have task specific agents by '26. Actually, I like that prediction because I can look back a year and say whether or not it happened. IDC says 1.3 trillion in spend by 2029. Who knows? I will say my view is this going to take a better part of a decade to play out because agentic is not just going to happen overnight, because of the data challenges and that's kind of your business. What are you seeing from customers?
Anand Raman
>> No, we're seeing that. And at the same time, I don't know if companies have 10 years to do this. So the conversation in every boardroom now is very different than what it was three years ago.
Dave Vellante
>> Sorry to interrupt, but to that point, I would agree. So you have to pick your spots where your data house is in order.
Anand Raman
>> Exactly.
Dave Vellante
>> You can't try to eat an elephant in one bite.
Anand Raman
>> Absolutely.
Dave Vellante
>> And so that's how it's going to play out.
Anand Raman
>> Absolutely. And so when I was talking about economics and experience, and what we are doing with Prism, to add to what Jeff was saying, the data problem, it is what it is today and you can keep improving it and it'll continue because the volume and velocity and veracity of data is not going to change. What we are doing with Prism now, enabling enterprises, is how do you change the SDLC experience for engineers and make what he was talking about predict and how do you proactively see what will happen, predict that, and make sure that it gets changed in the development process itself. So we've introduced a tool, something in the Prism context is something called SQL Maestro. What it does is it's actually a paradigm change from what many other tools are in the FinOps and operationalization space or observability space do. We are looking at all the SQLs that have been written that take the time and putting that in the context of how the business is running in your ecosystem, in your cloud ecosystem, and saying, "Just because this SQL is running in this context, this should take this and it should not take this." So it's proactively determining what to be done in the system and how the economics of the AWS infrastructure would play out so that somebody knows that, "Hey, this is okay to spend here and it's not okay to spend here."
Dave Vellante
>> It's a complicated matter because you're going from a software development lifecycle that was linear-
Anand Raman
>> Yes....
Dave Vellante
>> and deterministic to one that is non-linear and non-deterministic. So that's a whole different mindset.
Anand Raman
>> It is.
Dave Vellante
>> And so how did you guys respond to that? How are your customers adapting culturally to that? It's an education problem in part. It's obviously a technical challenge. It's a combination of factors that you have to then bring to market and help people adopt.
Jeff Veis
>> And one aspect of it is, if you look at the tools and utilities that are out there, they tend to be siloed. They tend to be product specific. There's a few guys that have things that will cut across maybe the hyperscaler platform, in which you look at two hyperscalers, so you can have that big view. Prism is the first AI powered observability offering that goes across the data platforms and the hyperscalers. So you can look at your entire stack and look at how that's performing, not just the pieces, which for an agentic AI deployment, it's essential. It's not a nice to have, it's a must to have. And that took some special engineering to be able to look across and not just report out how many terabytes did you push, how many queries did you run, but what is occurring between the data platforms and the foundation systems that are underneath them and look at them holistically. And that, we think, is going to be, that's the watershed event. And we feel that we're actually one step ahead, about a year ahead of the industry on that. That came from the work we do. It didn't come because we had some grand crystal ball. It came because as we're dealing with United Airlines, American Express, our major customers, they constantly came up to this issue of they're only looking at a piece of the elephant. They weren't looking at the whole elephant. That'll get you by for the traditional data processing, it won't work with agentic.
Dave Vellante
>> So you guys, I think you came on at theCUBE at Databricks, was it this year or?
Anand Raman
>> Yes. Yeah. I think this was early this year. Yeah.
Dave Vellante
>> Yeah, it was early this year. So Databricks got, Ali Ghodsi's vision toward open table formats and Iceberg kind of changed the game. Snowflake sort of, as you know, created the separation of compute from storage. Great. Okay, that was the modern data platform. And now that's evolving where any compute engine can act on any data And that's what customers want. But of course, it widens the aperture that you have to deal with. So Jeff, when you're talking about any data platform, you're talking about any data platform? Which data platforms? Databricks, Snowflake, Oracle, IBM?
Jeff Veis
>> The quick answer is yes.
Dave Vellante
>> Yeah. Okay.
Jeff Veis
>> We're strategic partners. AWS is one of our most important partner. We're a top 100 AWS premier partner. So that's why we're here. This conference and this relationship is very important. But we certainly do business with and partner very closely with the Databricks' of the world, the Microsofts of the world, Snowflakes of the world. And most of our customers have multiple pieces of that. We're not sitting there going, "We're going to be just an AWS house." And the agentic element of that pushes that out even more because those systems are going to be riding across it. So it was never an option for us to be having just a single vendor kind of solution. We had to look at the multi-vendor picture.
Dave Vellante
>> What do you think the success rates are going to be of agentic initiatives? Maybe I should flip it and say, are a lot of these going to fail? Is that okay? What do you guys think?
Anand Raman
>> I think it's okay for some of those to fail, and I think that'll happen. But I think the most important thing to take away is I think that the technology has promise, it has legs, and it is going to simplify and really allow enterprises to think differently about data and their processes and how they do conduct certain parts of their business.
Dave Vellante
>> Who do you guys sell to? And are there shifts to the personas in which you're targeting? Is it CIO? Is it developers? Are you selling more to the financial world, the line of business, chief AI officers? Who are your constituents?
Anand Raman
>> It's definitely not developers. We are selling to the largest of enterprises. So we have a very broad focus on the financial services side. Healthcare side, manufacturing. So these are three critical verticals that we focus on, but otherwise our customers are across many other industries too. So working with the largest of financial institutions, he talked about American Express and there are many others. We work with United Airlines on one of the airlines in the travel hospitality space. Work with the largest of healthcare insurance companies of the world, largest pharmas of the world. Those are our customers. We work with the CIOs, CIO organizations. We working with the CDOs. CIO and CDO organizations are deliberately being kept separate. And now there are many places where the AO office is also coming up. So CDO, CAO. So those are practically or typically our customers. And lines of businesses are there, but I think they are leveraging the CDO office and the CAO office to really facilitate what the line of business has to do.
Dave Vellante
>> Every time we get one of these waves, even when we don't have a wave, there's some new role. You remember the chief digital officer emerged and the chief data officer emerged. It's like our industry loves a vacuum, so we create a role. But of course, the chief AI officer now has emerged. And of course, it's like the chief internet officer back in the day. I don't even know if there was such a thing. But my point being, AI is going to be infused in everything. But you need actually a leader and a visionary who understands both the technology and the business aspects pretty deeply and then gets diffused into the organization. But the organizational changes right now are pretty dynamic. And again, it comes back to data.
Anand Raman
>> I'll add one more thing. It's not just about the data and the AI aspect. I think what's more critical now that an AI officer has to do is also the governance aspect of it, which is very, very critical. And in the world of where gen AI is going and agentic is going, I think the governance aspect is very critical for an organization because, while all the conversation in the boardroom is happening about that, but I think what the CIO is ultimately worried about or the CEO is ultimately worried about is, "Is my data secure? Is my data not going out?" Or, "I'm not believing an outlier and that's influencing the decision of an agentic."
Jeff Veis
>> And there's another part. It's remember we got dozens of LLMs. We have public LLMs that you want to, in theory, leverage all that wisdom that's out there, but you want to protect your IP. So your secret sauce is not now literally sitting out there to consume it. So these discussions of do I a RAG knowledge base internally? Where do I do that? How do I do that? And to what end? And think about the days of we got data, you got self-serve analytics, woo-hoo, I could do my own dashboard. And if somebody went off the deep end, maybe they make a couple extra data calls, okay? And the cost would go up and that would be the end of it. But now you have agents that may be talking to other agents, maybe actually resolving customer issues. We had one payments company, it was taking them up to nine to 12 weeks to resolve business issues in terms of payment issues coming up. It had to get escalated, it had to get reviewed and looked at. Very complicated because this was hitting their entire work streams. Now, AI agents, still with humans in the loop, I'm not saying those are out, they can do that in days and sometimes just in hours or minutes. And it's a stair step function what's occurring. But those actions that you're taking as you mess or correct your payment gateway, there's only one right answer. It has to work, right? Very, very different than the stakes that when we just had, if you will, next generation analytics. That if they were off, okay, maybe some report had some questions. Now, you are taking actions and you're in that path of risk as people are talking about today. So everybody has a higher stake. They have a requirement to do it to compete, but it's got to work. And the thing is, it does go back to the foundation underneath it. If you cheat, or try to cheat, and not have a modern foundation that you can trust, that you know the data's good accessing the right way, that has the performance you need, that metaphor of you're only good as the foundation you're building your skyscraper on. So you don't get to like turn the page and forget about that stuff. We have clients, they're still running on TD, they're still running on mainframes, especially in financial services.
Dave Vellante
>> Yeah, of course.
Jeff Veis
>> So that's still there plus plus.
Dave Vellante
>> Yeah, we got to go, but it's ironic. You've got these countervailing forces. People want open data, but governance is the number one issue, right?
Anand Raman
>> Yes.
Dave Vellante
>> And so it's a complicated matter. You've got the pressure to trial these LLMs and experiment and actually find value, and, at the same time, be safe. So, and again, you guys doing some good work. Appreciate you coming on.
Anand Raman
>> Thank you very much, David. Good to see you.
Jeff Veis
>> And if I can give one plug. If any of your audience goes to Impetus.com, you can go right to the Prism site and there's an Experience Center where you can either take some sample workflows and data or you can push up your own. And in just 15 minutes, you can actually see what this native AI driven interface does and what kind of observability it's going to give you in the ways we talked about.
Dave Vellante
>> Pretty cool.
Jeff Veis
>> And people should check that out.
Dave Vellante
>> Yeah.
Jeff Veis
>> Don't watch Stranger Things. Do this instead.
Dave Vellante
>> Guys, thanks so much for coming on theCUBE.
Anand Raman
>> Thank you very much, Dave. We appreciate it.
Dave Vellante
>> Good to see you again.
Jeff Veis
>> Thank you.
Dave Vellante
>> All right, keep it right there. This is Dave Vellante, for the whole CUBE team. We're live from Las Vegas, re:Invent 2025. We'll be right back right after this short break.