theCUBE + NYSE Wired: Mixture of Experts Series | Nachiket Deshpande, Impetus Technologies

Clips
More from theCUBE + NYSE Wired: Mixture of Experts Series

Nachiket Deshpande

CEO

Impetus Technologies

play_circle_outline LEAP AI: Liberating Legacy Data with LEAP Logic and Semantic Context Fabric for Prism-Governed Agent Solutions

play_circle_outline Airline lost-baggage use case: context engineering reduced lost-baggage rates via tailored agent solution

play_circle_outline Combined software-plus-services model: one-to-one customized product with outcome accountability

play_circle_outline Encapsulating Complexity: A Platform for Databricks, Snowflake, Hyperscaler Integrations and Rapid 6–8 Week Delivery

play_circle_outline Observability Control Plane: Managing Agents, Measuring ROI, and Adoption Paths

play_circle_outline Goal: scale production-grade agent use cases across customers to deliver ROI

Info
Transcript

Nachiket Deshpande, Impetus Technologies

Nachiket Deshpande

CEO Impetus Technologies

In this interview from theCUBE + NYSE Wired: Future of Agents series, Nachiket Deshpande, chief executive officer of Impetus Technologies, joins theCUBE's John Furrier to discuss the launch of the LEAP AI family of products and how context engineering can close the gap preventing enterprise ROI on agentic AI. Deshpande walks through the four components of the LEAP suite — LEAP Logic for data modernization, Context Fabric for converting structured and unstructured enterprise data into actionable context via knowledge graph ontology, an Agent AI Solutions envir... Read more

explore Keep Exploring

What is the newly launched LEAP AI family, what enterprise problems does it address, and what are its main components and governance features? add

How can a contextual/agent-based data solution be used to reduce an airline's lost-baggage rate (for example, from 10 per 10,000 toward 7 per 10,000), and what were the results in a real-world case? add

How does the new approach combine software and services to provide a customized, outcome-oriented solution compared with traditional system integrators and standalone software vendors? add

How does the LEAP AI suite help solution builders and customers stay current with rapidly evolving AI models, APIs, and data ecosystem technologies without freezing their architecture? add

How do you expect the Leap suite to be adopted and rolled out within customer organizations — what are the typical entry points (data, use-case, observability) and how does adoption progress over time? add

What would success look like for your company over the next year in terms of customers deploying production‑grade intelligent/agent systems and generating measurable ROI? add

bolt Powered by CUBE AI

Nachiket Deshpande, Impetus Technologies

search

John Furrier

>> Welcome back around to the Cube. I'm John Fur, your host here at our NYSE Studio. Of course we our Palo Alto Studio connecting Silicon Valley and Wall Street. Got some big launch news here with a bit of technology. We've got the CEO NCA dash pane back on the cube. Big news for your company. We've talked previously about the, the macroeconomic conditions and the tech scene, how the business model transformation are happening. Let's get into the big reveal LEAP AI solutions, family of products building on the platform shift. Let's get into the news. Explain.

Nachiket Deshpande

>> Thanks. Thanks John for having us here. Really excited about this new launch. Last time when we spoke we talked about the sort of context gap that is sort of preventing the ROI realization for enterprises on their AI journey. And we are really excited to launch the LEAP AI family to essentially bridge that gap. We talked about the data gap being one key pillar, why the models do not have the enterprise knowledge that enterprises are, which is unique to their enterprises. So we have under leap.ai, we have LEAP Logic, which is our modernization solution that helps you liberate that data from legacy environments and make it accessible to your agent AI systems. The second component of our lead I suite is the semantic layer with knowledge graph ontology that essentially converts both structured and unstructured data into the context that is needed for the agent AI activation and helps you navigate that context through the systems to make it available. The third component is the agent AI solutions environment where we are actually able to bring all these kits together with skills and context to solve a specific business problem and there is a governance layer through Prism, which brings in observability, guardrails and security engineered right around every step because that also is becoming a very critical element in agent care adoption. So that's the leap suite for you. We are really excited with the launch. We've been working with number of our existing customers to market test it and and see the relevance of that excited with the initial results that we have seen. So really looking forward to taking that to the larger enterprises around the world.

John Furrier

>> Yeah, we talked last time and this is again an industry theme on almost every keynote from I've heard from CEOs on stage is closing the gap. Everyone wants to close the gap. You talk about the context gap. Talk about why this leap ai, I love the name by the way, leap leaping forward. It's very, very visual. Talk about that gap that you're closing. What specifically are you addressing in the leap here? Because what gap are you closing? What specific problem are you targeting? Obviously context, we talked last time, this makes total sense as the top conversation. What is the core sweet spot of the target here?

Nachiket Deshpande

>> So maybe it'll be useful if I give you an example. So one of the, you know, top airlines in the world, it's been our customer for the last more than 15 years and we've been working with them on their enterprise data journey along the way. So the, the, the chief data officer of that airline, he sort of threw a problem at me when we were talking about all of these agent K use cases and potential. And his problem was very simple saying that as an airline today we have a lost baggage ratio of, I'm just taking the numbers as example 10 in 10,000 and that 10 in 10,000 white might look great on percentage terms for the airline of their size. It's several millions of dollars that it costs in order to route that back back to the passenger or compensation that they have to pay for the delays or losing that back. So their challenge from their business was huge identity ki to help me get it from 10 bags per 10,000 to seven bags per 10,000. And now that problem, they were trying to solve it with models for a while and they were not getting anywhere with that. It was going only up to 80% accuracy and hence it was not good enough for the business to adopt into production. So when that problem was shared with me, we actually came up with this context solution where we said, you know, what are the data elements that influence a bag, making it on the right flight and how do we model that around that? And then when we built the, we layered the data that was sort of buried in the enterprise systems for that airline also figured out how do they, what are the priorities that they assign when they look at baggage transfers, what is important for them sometimes transitioning of aircraft from the bridge that SLA is more important. Yes, certain airports, they have their own staff, certain airports they don't, they partner with someone. All of these are very specific airline nuances. So we brought that context in and then we are able to now build an agent solutions that is bringing it down from 10 to nine to eight and a half and we'll continue to fine tune it to go get to that goal. That's the example of what the LEAP Suite can do for you.

John Furrier

>> I love that example because what you just said is consistent with some of the things we've been talking about in the Cuban we've been seeing in the industry where it used to be the users and the use cases bend to the technology. Now the technology bends to the use cases. You mentioned that, yeah. Example, many use cases per the airlines. I mean there's there's different situations context, correct. That's and the execution workflow. Yeah. So that's an example. So I love that on the launch you guys got the modernization layer with Leap Logic, you got the semantic layer with Context Fabric, you got the observability with Prism and you got the agent solutions piece which builds on the services. So I have to ask you, how did you get here? Because the difference between Genius is timing, you know, it's, you've had the peace parts and the timing of this is in is consistent with good architecture, the governance focus built in from day one to having that systems play. How, how'd you get here? What was the timing internally? Can you share your thoughts because this launch and rebrand is consistent with what people are talking about is where they're thinking about for agents. Yeah,

Nachiket Deshpande

>> So there were components that we were sort of already at it, right? So LEAP Logic has been around for some time, which we have been, it was our key platform for modernization and we were doing these journeys right from sort of big data to data on cloud to lakehouse to data for ai, right? So we've been through this journey with enterprises. That was always our strength. We built upon that strength. The piece then and observability part also was always there because we built it originally for finops in the cloud world. And so we had the capability which was for the different purpose. But as we started to talk to the customer and understand this context gap and the solution for that, the, the agent solutions and the context fabric were the components that got visualized as the critical missing pieces to make this platform a holistic platform for agentic transformation. So it's the same thing that we talked about earlier is it wasn't a pivot but it was an inflection point for us. We took the strengths that we always had in the logic and in Prism we accelerated on molded them for the agent world and then invested in the context fabric and agent solution to make it complete platform so that we are able to get the ROI problem fixed for enterprises as they look at their agent journey. How,

John Furrier

>> How should people understand your approach? It's a new approach, it's a family, it's a platform. What's the new approach? How would you describe it?

Nachiket Deshpande

>> I think it's a new approach that traditionally you would've seen system integrators which will build a lot of knowledge and capability to the table and say I can solve all of these, but it was bespoke and it would've taken a long time to get the resolution done and the outcome was not always guaranteed. And on the other spectrum you would have software companies who will say, I have a solution, here is the outcome. But that solution is not tuned or not customized to your needs and hence you will get that piecemeal outcome but it won't solve the end-to-end problem because other pieces around it are missing. And then the system integrators played a role to sort of stitch these, those, these three tools, two things together new, our new approach is that we are bringing the software and services together on it. So it gives you the benefit of outcome area orientation that the software is used to do. And it brings in the flexibility of the services approach that used to do so. Essentially think about it as one-to-one customized product. That's really how we think about it. Where the product is absolutely adopted to your needs, your environment, your enterprise, but we are taking the accountability for the outcome with you like a software word. And that combination is what makes the impetus approach unique.

John Furrier

>> So your platform's redefining the category of on-premise services with partner integration firms coming together. I wanna get your thoughts on this. I think this is really kind of clever because the success use case that we've been pointing out on the cube has been company's readiness with the velocity of AI capabilities. I mean everyone's talking about, hey in the next three 30 days something else happens. So there's a lot of velocity on the AI capabilities. So the readiness is really where the companies are more successful when they're ready, timing, execution, how do you address that? Because we're seeing a lot more co-design going on between integrators, whether they're medium, small, medium large, or the large GSIs that have traditionally been like bigger projects slower, but there's a speed mandate. Yeah. Talk about that dynamic and how this fits into that.

Nachiket Deshpande

>> So the lead AI suite is kind of doing sort of two things for that. Number one, it is encapsulating complexity of underlying technology changes from the solution builders so that they are able to keep up in a consistent way of building solutions and not worry about a new no model being launched or a new offering being launched or a new API being launched. The the platform encapsulate that and allows you to adopt it progressively. And the second thing is that we are building this in collaboration with our key partners. So we are also in step with them how the technology is evolving. Whether these are data ecosystem partners like the Databricks or the Snowflakes or hyperscalers or model partners. We are working with all of them in ensuring that we are able to bringing newer capability seamlessly through this platform to our customers. So we are encapsulating the complexity but ensuring that they continue to get the leverage on the new technologies that are coming in. And the third piece is when we do software plus services together, it really shrinks the cycle. So we are doing most of these as six to eight week cycles. So in a way, even within that cycle, if new technology comes in, you have an opportunity to catch it in your next sprint. You're not really missing out on that, you're not freezing your architecture for a long time. And then hence worried that whether by the time you're done will it make it relevant or not because it's the speed and the encapsulation of that complexity actually solves for this problem through LEAP AI suite.

John Furrier

>> Well it's a great, great rebranding, love the Leap name. Based on your experience, you've seen a lot of projects. How do you, what's your vision on how you see this unfolding? Take us through your mind's eye around, okay, fast forward it gets being deployed, there's adoption. What are some of the things that are gonna come out of this? Because all the successful AI projects are producing a lot of value fast. The extraction of that value we're seeing not just cost reduction but revenue generating. There's not a lot more ROI opportunities to score points to build that equation. What do you see unfolding if you, if you had to kind of peek into the crystal ball, I mean ideally in a steady state, how would you see it unfolding? What would be a day in the life of this rolling out into a company?

Nachiket Deshpande

>> So we typically see your sort of three entry points of this leap suite into our customers. There are some customers which are ahead in this adoption journey, have fairly good discipline data discipline within their organization in those customers I see it enter through the data lens and progress through the context journey as they start to make their AI adoption mainstream. There are customers like the example I gave you about the airline where we will enter with a particular use case that we want to solve and we will sort of move backwards and integrate through context and into the data relevant for solving that problem. And in some customers we are entering it through the observability lens where they've built a lot of solution but they don't know how to manage those because they've sprawled in different teams and they have now certain have X number of agents in production. They don't have a view of how do I, how do I manage that? How do I observe, how do I really make sense out of that There the entry point is through observability. So if I look at fast forward we would see sort of different angles of adoptions of that platform and eventually sort of start from data, go into the agents, start from agents, go into the data or we see the observability, which is, you know the point that we mentioned that they have a lot of agents but they don't know how to manage it. So we will, we become that control plane for them to be able to manage the environment that is kind of getting out of control a little bit through. Yeah, this pieces.

John Furrier

>> Well I like how you guys had the piece parts, the market timing was good. I mentioned that timing piece is critical. Obviously there's demand, a lot more work to be done. So a lot of headroom. Can you share the company's beachhead that you've established? Because how you got here was through a lot of work building the technology. Now you got the family of LEAP AI services and solutions, you had business going on. This is an extension to that. Yes. You've done a lot of services business, a lot of migrations. You've been in the, in the weeds a lot with a lot of customers. Now you have this business. So it seems to me that you're in the context business. Yep. Not the services migration business. Share your vision on this, this change inflection of the company and the importance of that.

Nachiket Deshpande

>> Yeah, so I think one of the main sort of inflection point now is earlier we would get engaged with the customers when they would've decided to modernize their data states. And we would be a key partner to help them do that. And, and, and I would say that we did it better than anybody else but we got engaged when they had made that decision. Today we are actually talking to them on why are you modernizing the data? Yeah. And what's your objective with the modernization? What's the ROI? How would you actually see that helping your customers, your business? How would A CFO see value in it? How would a business head see value in it? And we are becoming the partner to actually make that ROI happen for them. So we are, you know, starting point in our discussion with our customers is materially different and we are helping enabling them to achieve that ROIA lot better than we were when they had decided what they wanted to do. So that's really the fundamental sort of things changing and hence we are leading with context. We are leading with context in every single conversation where the teams is sort of also being trained to ask, okay, but why are you doing it? Yeah. Why do you want that? Why do you want that? Till we get to a point where we understand how the ROI for that customer will work. Yeah. And we want to be the partner to get to that outcome. So

John Furrier

>> You guys built the muscle. Absolutely. And you have a trajectory with economies of scale in that service piece. Correct. That gave you a front row seat to build the requirements Again, timing Yeah. Is back to this. Yeah. So would it be safe to say that you're in the context creation business? 'cause you have the governance, I mean everyone's talking about governance. I don't think, I think I've said the word governance more in the past year than 17 years in the cube outside of certain, you know, air, certain verticals. But everyone's talking about governance is the key to getting those rules for the agents. Yep. I mean context creation, that's the business.

Nachiket Deshpande

>> So I would probably speak a little, okay, my,

John Furrier

>> You're the ceo - So we are in the context engineering business. Okay. And I, I would differentiate between creation because creation would imply it's a one time activity. Yeah. Unfortunately it's not, you know, by the time you build the context underlying systems and data change every day, every minute in some cases. So you have to, it's a living organism within an organization. So you've got to, so you have to have systems thinking for context. Yeah. So that it is engineered, which means that it will not only it's created, but it's able to keep it itself relevant all the time. Okay.

John Furrier

>> That's good distinction. That's, well that's why you're the CEOI would say that the data changes in context. So the context engineering is this platform piece to the business challenge of 'cause more data's coming in. Correct. Okay. So when you look at the future, where do you hope this will be in a year? What do you hope the outcome will be from this? The type of engagements? What use cases? 'cause agent's coming on super fast. We're about a year, in my opinion, away from full explosion of agent agentic action. We're seeing use cases being, you know, using agents. Now people are being practical and intentional on the agents. Where do you see this evolving on the agent world?

Nachiket Deshpande

>> So I think if you, if we sort of, we get this right, I believe that a year from now we would have lot more customers with production grade intelligent systems than we see today. And, and I'll sort of relate to that in the past as well, is that when we were in the modernization journey, we always took pride in us being able to help customer decommission their technology more than anybody else. Right? That's the outcome orientation that always differentiated imp peter us. Yeah. If I take that outcome orientation into this new world, for us it's less about adoption of my platform, but more about can I have a large cus portion of my customers have significant agent use cases in production that are scaling and delivering ROI for them. That's the success for me today. That percentage is 2, 3, 4, 5. We would want to get 50, 60, 70% of the use cases that our customers are interested in into production generating ROI that would be the success for

John Furrier

>> Us. Question on the enterprise adoption and usage. The value I get, I get that that's clear. A lot of enterprise architects are thinking generational decision when they think about, you know, horizontal scalable control planes, semantic layers, harmonization layers, all the things we we do in the deep tech. How should they think about you guys on the engagement side? Is it I need that this, I need this everywhere. What levels of to, can they put their toe in the water? How do they adopt? What's the engagement program and what's the 'cause it's a heterogeneous market. Yes.

Nachiket Deshpande

>> So the way we've built our platform is actually, that's kind of kind of why we call it platform because it's, it's, it's plug and play. It is scalable and it is customizable to a specific environment. So number one, we would, we would adopt to some of the technology choices that our customers might have already invested in. And we're not saying that you throw away everything you have and get something new because that's, that is always a longer

John Furrier

>> Not a rip and replace. It's

Nachiket Deshpande

>> Not a rip and replace. It sits very well within their ecosystem. Also, the platform abstraction, sorry, abstraction layer in the platform is actually meant for reducing the stakes in those decisions because people think that I have to decide what model to use. Yeah. And to my approach to that is, any decision that you take like that you're going to have a buyer's remorse because tomorrow morning some other model is going to come in, which is gonna be powerful with the one that you invested in. And you're gonna say, you know, did I take my decision too early or, and was I wrong in my decision? Yeah. So you have to encapsulate or abstract it in such a way that you should be able to continuously take advantage of technologies as it evolves. So hence your decision of some of those technologies. Not a generational decision. Yeah. But it's a, a very purpose built tactical decisions you're taking in point in time for a specific purpose. So those are the two things that we, I believe will help customers adopt it progressively and reduce the stake in the decisions that they're making so that they can continue to remain current on those.

John Furrier

>> Thanks for coming on the cube. I guess final point is, how would you describe the, the news and launch simply to someone watching? What does it mean, what does it do for them? What is it? How does it work?

Nachiket Deshpande

>> So every single enterprise which is struggling to justify ROI on their agent AI journey, leap AI solutions and our context engineering methodology will help them solve that problem.

John Furrier

>> Yeah. Closing the gap, accelerate the value, of course, we're here in the queue, bringing that leaping ahead leap AI platform. This is Jennifer in the cube, thanks for watching.