Panel · 11:55 AM · Wednesday, May 13, 2026
All right, turning bets into business results. This is going to be probably the best session here. I expect in your reviews to see nothing but fives, or if it's a ten scale, nothing but tens. Because that's what we care about at the end of the day is how well did Joel do? No, I'm very honored to bring together five folks. A great variety here. We've got Steve. Steve, I've known for a while, distinguished KPMG leader, one of our board of advisors. Rahul, I'll just go down the list here, chief information officer, very large bank, dealing with AI on a daily basis, setting the strategy. Josh, also on the enterprise side, Healthcare, public sector, dealing with a ton of things. Some of the conversations already came up in Phil's panel that we're going to build on. Aravind, global leader for UST. We have a lot of conversations, especially around one of our passions, which is telecommunications and how that networks are modernizing everything, especially in the age of AI. And then Eric. Eric's got a really interesting point of view from his role at Hashgraph around emerging technologies. And fundamentally, we tend to forget there were past emerging technologies and how we navigated them. So we really want to kind of don't forget the past. Learn from it and execute. So that's going to be critical to how we turn those AI bets into real fiscal, manageable, tangible results in our business. So this is going to be a quick series of questions. Really looking for a dialogue amongst the folks here. And ideally, right after when it's lunch, these guys aren't going to eat. You guys are going to be so interested in talking to them that you're going to want to basically pull them away and get some of that knowledge and insight that they shared up here and expand on what that matters for you and how you can go deploy that. So let me start with this. I'm going to start the first question to you. Right. What is the next domain you're looking at to scale AI? And what is the criteria that you apply to making those decisions?
Sure. So, of course, as a financial institution, it should be the customer experience specific AI use cases. However, you know, as a European bank, EU AI Act kind of stifling the innovation. You know, you have to be careful. So, yeah, still we are putting effort and implementing certain AI use cases. Now when we are talking about AI, always everybody's about the Gen AI, talking about Gen AI. But otherwise, we have been implementing AI, the traditional AI machine learning, and we have a big team for implementing those kind of use cases. But now it's coming back to the Gen AI, and that's where we have to be careful with the EU AI Act and the privacy issues and all different kind of issues we have to worry about. Now coming back to the next domains, I think in order to foster the digitization, and that's any bank or the financial institution have like digitization, the customer experience is the priority. So how do we enable that? Yes, everybody's talking about agentic AI and AI use cases. So in order to enable that, where we would go for? So one would be, and I think Manish brought up, like the developers are not that many in a bigger scheme of things, but in the develop, like if we can expedite or increase the speed, speeding up the development and reduce the development project cycle, that would be really helpful and useful because that will enable further digitization effort and then AI use cases too. So that's the next domain. And that's also easy to explain or convince. Because now, and then also compared to other use cases, I think the measuring of ROI or the KPIs are there. If you do properly. Okay, you know, you don't just count the development improvement or the speed. But when you measure the entire development project, you will see the differences and whether there is improvement. So that's the good news. So that's where the focus should be. And then now there are tools and good maturity is coming that would help developing or generating the code. So that's one place and that's how that can be convinced and can be properly measured. And the second use case and easy to, and then actually it's a really requirement now for the financial institution would be the cybersecurity. And we have been talking about all this AI-enabled attacks, cyber security attack, and yes, indeed, in the past, there were script kiddies, and they were initiating the attacks, and then their statistic was in our side, because generally you would be surviving, because not, there are certain actors, they would be focusing and then attacking you, otherwise, you know, statistically, all those script kitties or those hackers will be going after you. But now with the AI, you know, you have, you know, easy to, they can go after many organizations. So now we need to protect using AI. And there are a number of use cases that organization can work for. One could be the continuous penetration testing and the vulnerability assessment. The way attacker could use AI to penetrate, we can do too and find out whether there are vulnerabilities. There is a way to breach in. So that's where, you know, one good use case. The other one would be the threat hunting. Again, there, I mean, a lot of manual effort right now applied, used different tools. Now that agent can be trained to complete the threat hunting, and there are some other use cases. So again, to me, the development part, the cyber security part would be the two domains we are easy to convince the board or whoever the decision maker. And immediately you may gain some benefit out of it. Other use cases like, okay, internal operational improvement. This is like something was discussed before for RPA. You know, like now it's an agentic AI. But the underlying problem for RPA adoption and agentic AI remains same.
Good. That's interesting looking at it from a CIO's level. Eric, as a CEO, looking at your business and looking across the divisions in marketing, sales, technology, operations, what is your criteria for your teams coming to you and saying, hey, we think this is a good place to put our bets on AI?
I realize we are actually competing with AI right now in the room because I see people checking their code and their ChatGPT on their phone. But it's actually the other way around, right? So when we started the AI journey at Hashgraph, we actually said, as an executive team, we said we're going to embrace AI. It was a year and a half ago. And we're going to push AI on everybody in the company. So it didn't come from them to me. It's actually a decision we made at the executive level. And we are 200 people company. So it's much easier for me to do this than for many of you, I realize. I used to work for IBM and for Deloitte, so I know the pain. But I think that's how we started. And then obviously the tech people were the first one to embrace. So the engineering team, the product team, the cyber security team to where I was pointing, they embraced the adoption of AI in particular very quickly. And we've seen tremendous impact on the roadmap of the product. So we announced something really interesting last week, if you're in the blockchain. How many of you were in the blockchain ecosystem for the last year or more? Not many of you, but if you're in blockchain, what we announced last week is awesome. It's called CLPR, but that was my commercial break. And we actually were able to deliver CLPR last month because of AI. We were planning on delivering in March of 2027, and we delivered more than a year in advance. That's an example of where AI may make a huge impact on the startup like us. And I think some other teams are struggling to adopt, right? Let's be honest. I think HR is struggling, right? Because it's a people business. So they want to be close to the individuals. They want to connect. And letting go of that connection is actually hard for an HR function. So you have different level of adoption. But I think we all made the decision at the same time, which is, I think, very important.
You've both kind of opened up a nice, easy segue to the next question. Looking at what the CIO is looking at, what you're looking at, and complementing that. Josh, as a CIO yourself, how do you see KPIs changing? Because at the end of the day, we're enabling people with these AI devices. The KPI that Eric just spoke about with regards to bringing a product to market faster. How are you and your organization seeing KPIs change?
I think the short answer is what we've already heard from a number of panelists. It's going to be about the outcomes and not going to be about how many tokens or whatever that people are using. But I'm going to apologize to the audience because I'm going to get into the monstrosity that is the U.S. health care system. Independent Health is a not-for-profit community-based health payer. That means we provide insurance. We're based in Buffalo, New York, and all of our membership is Western New York. What we found is that it's nearly impossible for us to maintain a zero margin. Why? Because we're constantly losing money year after year after year. Why? Because the cost of health care is so high. So I'm getting into this because I'm going to show what we're thinking about from an AI use case perspective. And I'd love over lunch if people want to talk to me about these things. But I'll tell you how we're thinking about it right now. Last year, we implemented a program where we looked at the sickest of the sick amongst our population who are members of independent health. And those members are driving 80% of our cost. Why? Because they're going to hospital all the time, going to emergency room visits. They don't have consistent access to care. So what we thought we could do was intervene in their journey, in their medical journey, by sending people to their homes. And we offered this as a program, and we offered it for free to the members because the thought was that for every one of those members, if we could cut their costs maybe even in half, it would ultimately save more than the cost of this program. And the good news is it did. It saved quite a bit, but generally these members are still a cost to us. So that sounds like a success story, right? Actually, it wasn't. It was a disaster. How can it be a disaster? Well, the next year, what ended up happening is the same people re-enrolled in the plan, meaning they brought the same costs in, and we had to again apply this same program to them, which is still at a cost, right? So we're not making, we're not turning people profitable, so to speak. We're just continuing to drive cost. And all of them told their friends how great this program was. And so now, all of a sudden, we actually doubled our costs. So what can we do differently? And this is where AI comes in. One, can we enable the nurses who visit these patients while they're driving to the patients, can they be on the phone with an AI learning about what the patient's needs are that they're visiting so that they're ready when they get there to help that patient immediately. Because what was happening was they would get there, then learn about the patient while they're in front of the patient, and then go see the next patient. So if we can use AI to help that nurse interact on the way, that means the nurse can see more patients in a single day. We'll get more people through this program. There's three total ways. The second way are the people who decide what patients these nurses should see. They're looking at dashboards that we're producing for them. I'd rather they produce their own dashboards that might change on any given day based on what they're thinking about and what they're seeing in the community. So what can AI do to enable these positions to basically converse with their dashboard to identify who the patients are for those nurses to see. And then the machine learning side, the deep detailed algorithms that we're producing to determine who exactly are the patients that fall into the program to begin with. And so each of those, you can see, has its own collection of outcomes as opposed to, well, how many tokens did we use on that day?
Aravind, what are you seeing in terms of KPIs when you're talking to your clients?
Yeah, I mean, you know, we're actually seeing things both from an internal perspective as well as an external perspective. Because, I mean, UST, we are a digital services provider. So we're seeing Fortune 1000 customers. We've been in business for about 27 years. And we work with a select group of customers. We go very deep. And that's our strength. And what we are actually seeing externally from a KPI perspective is that, you know, when you actually make the KPI the target, people figure out ways to game it, right? And that was called out in Phil's keynote as well, the Goodhart's Law. We saw that happening externally as well as internally when we actually started measuring tokens. Because a couple of years back, we actually started out with, how do we encourage our folks, 30,000 of our folks, across multiple continents to use AI? And there were training programs that we put together. We brought in Claude. We had OpenAI, best of the tools. But the next question was, are people really using it? Let's actually measure AI tokens. And that was not working because it was working in the favor of the LLM providers because the bills were actually going up. But we were not really getting those results. And we actually see the same thing from a customer perspective as well. And going back to what Joshua said, we quickly realized that while those KPIs are really key performance indicators, not the absolute truth, you need to actually layer that out with real business outcomes. So we shifted towards, you know, I mean, it's so easy to actually get caught in what's the dollar savings that we can actually deliver for our customers. And those are things that we have been actually doing for our customers from an automation perspective, you know, having lean teams, etc. But we really started shifting towards how do we actually generate revenue for our customers. I mean, case in point was, you know, I mean, telco is close to my heart. We actually did a project for one of the tier one telcos where we had this treasure trove of data being generated out of their radio access network, right? And, you know, I mean, the first problem was how do we actually manage all of that particular data? But we go through that. And we actually started generating models that would actually predict when a network equipment is going to actually fail. And that was good. And it was actually reducing truck rolls. There were cost savings. When magic started happening, when we actually started predicting in which market a radio access network is going to go down, and connecting that to customer churn. And having the promotions team put a promotion into that particular market, now suddenly we are shifting it from a cost optimization perspective into a revenue generation model for the customer. So those are the things that we are actually seeing that is scaling. So, I mean, from an organization perspective, we are actually focused more on how do we actually generate more revenue for our customers, shifting away from how do we reduce cost, right? I mean, that I think is the kiss of death for service providers like us. So that is one thing. And the other element that we are also seeing is, and we have had good examples in the past, where we actually see intersections, interesting intersections that we can make possible because technology is there. Intersections where IT and OT comes close together, right? You know, I saw other panels actually talk about process complexities, et cetera. That is an element to be addressed, right? I mean, I don't think we are actually giving enough importance to that process re-engineering or, you know, killing out the processes and coming out with new elements. I think that process re-engineering is going to have a moment, and that's an area that we are actually focused on. And bringing in AI, right? So these are some of the elements that we are seeing and how we are actually addressing that.
One thing we were talking about in the green room before we came out here was how often these emerging technologies blind us to what we learned from the past S-curve. Eric, I just want to start this conversation, then I want everybody to chime in on, what did you learn from the past S-curves that you were involved in that we should not forget and we should apply to AI? So, Eric, what do you think? What was the, you talked about blockchain. Now we're in the AI world, in this early stage of the S-Curve. What do you think people are in danger of forgetting that they should really say, no, we experienced that, we learned how to handle that, just evolve our thinking?
I'm gonna be embarrassed answering the question, right? But that's part of the thing. We went through the adoption of the internet, the adoption of cloud, the adoption of mobile, the adoption of a lot of different innovations in the last, what, 50, 60 years, and more before that. I think this one is a little bit different because this one is much faster in terms of change than the one before. We were talking backstage about the S-curve. The S-curve here is very compressed compared to where we were. But lessons learned can be also extracted. And I think the first one for me was, I've been in blockchain for 13 years. Some people say I'm crazy, right? You should have given up a long time ago. But I think the change management and the focus on real business output is the key. We've been waiting for a killer app in blockchain for 10 years, and finally we got one. It's called stablecoins, if you follow a little bit what's happening. But it takes time to get there, and in the meantime we spend a lot of energy and time on doing things that really didn't matter. And so I think you have to be patient, you have to be resilient, but you also have to focus on the business outcome. For the last few years now I've been saying no one buys blockchain. No one buys blockchain. There is nothing to buy. You don't buy blockchain, you buy an outcome. And I think in AI it's the same. You can buy tokens, you can buy AI, but that doesn't matter. What matters is what are you gonna deliver to your clients, to your members, to your customers. That's what matters the most.
Josh, what about you? What did you learn that you've forgotten since the AI rush has come up that you would encourage people here, don't do what I did?
Yeah, it's not that we forgot it, we just ignored it. We've always known to try to avoid vendor lock-in, to take your time, do the appropriate due diligence on the contracting side, and know that the decisions we make now could potentially end up being technical debt later. I'll just give another example, once again, of something that appeared to be a success and is now becoming a challenge for us. I swear that's not everything I do. There are things that were a success and continue to be. We implemented an AI where if you call our number, you could look up our independent health on the web and dial our number during business hours. An AI named Rose will answer, and she'll authenticate you, and then route you to an agent. And we did that because we had the largest open enrollment in our history when people subscribed to our services, and we knew that was coming, and instead of hiring dozens and dozens and dozens of temps, we implemented this AI. And it worked great. It worked great, and we got the ROI. But the company that we used for that technology, they're now saying you didn't implement it using an agnostic LLM, you used a specific LLM model that we no longer support. And you're going to need to upgrade, and by the way, that's a huge cost. So that was one, and that's the technical debt example, and the other is the phone system that we integrated with, they're now saying, hey, you should be using our native AI, which, by the way, is terrible. You should be using our native AI, and because you're not, now there's a cost. There's a line item that you probably didn't notice on your contract, and in the next renewal, that's going to be 10 times greater. Or you could use our native AI for a lot less, but it is terrible. So, you know, these are the things, right? These are the things we kind of knew about, but now it's coming back to haunt us.
Rahul, examples of your forgotten choices and your new journey?
Not forgotten, but I think, and I alluded before as well, I think from the history, from the past, one has to learn. AI has a potential, that's why all of us are here. But I think Manish mentioned before that only people are making money are model providers right now. Otherwise there is a big chaos and all of us do realize that eventually this AI baby is going to grow up and going to mature and everybody's going to reap the benefit. But one has to be careful. I think several years ago there was RPA bandwagon too for the internal process improvement and all. Let me ask you here, you don't have to answer, but ask yourself how successful all these RPA projects were. So, and then why they failed. So those underlying issues, still for the agentic AI, yeah, is another bandwagon, and it does have good potential. But it has all underlying issue and many more. So that has to be settled, that has to be fixed. Then yes, once you learn all this thing, then yeah, eventually there is a calm water and everybody's feeling good. But also one other aspect, which there was a woman here, I was other side sitting, and she did talk about, you know, the workforce. The goal shouldn't be the reduction of the workforce. Organization have to be careful about the change management, how you are sending the message out. What's the purpose of this AI adoption? And these are the ones, something we have to be careful and learn from the past as well. And that's how we slowly will adapt. So to me, it's still hype. And so we are not there yet.
Great point. Aravind, I'm gonna skip you. I'm gonna jump to Steve for a second here because I've kept him quiet for too long. Because one thing that's kept up, or has come up several times in this conversation, is speed and security. And one of the conversations Steve and I have had in the past is the changing clock speed of things. Steve, what are your thoughts, as you're advising companies, as you're helping companies develop their strategies, how are you advising them the impact of speed on everything that we've been talking about? Where does cybersecurity fit into that?
So I appreciate you asking me. The fact that I've lived through a lot of these S-curves, from client-server adoption to the web to cloud to, you know, e-commerce to now we're living in an age of AI. And what I would suggest is that people oftentimes focus on speed the wrong way. The issue isn't generally speed, because people will tend to look at speed as a KPI, going back to KPIs, in a specific area. They'll tend to silo the speed dynamic. The real issue is clock speed for the organization. Every one of these technology S-curves has increased the requirement for clock speed in an organization, starting at the board, all the way through. And the key metric that people don't pay enough attention to, people say people, process, and technology. People, process, and technology has been part of all these S-curves. The piece that has missed, oftentimes, is governance. Governance sets the pace for people, process, and technology. And if you don't approach it, the board has to initiate this oversight to management to say, how are we evolving our governance to increase our clock speed as an organization, to take advantages of the opportunities by these new technologies? And AI, by the way, is I think it's a loud amalgamation of the so what, because the two dynamics that AI brings to bear are a little different than what we've seen before. The first dynamic and we mentioned this I think Phil mentioned this in the outside was that AI doesn't know boundaries. If you let an agent go, it'll go where it needs to. No technology before has ever done that. We've always sort of constrained through ex-software engineers like myself the boundaries of where technology can work. AI has changed that dynamic, that's important. Secondly, the people using AI today are different from the people that waited to be provisioned for IT support in the past. Our employees are getting AI'd whether they like it or not. They're coming to work with these things. And this is powerful AI. Those two dynamics change the speed and clock speed dynamic for today. And I think governance is a very, very underplayed attribute of important success in this S-curve.
Aravind, you're nodding a lot. You want to build on that?
Yeah, I mean, I want to actually go back to what Rahul was mentioning about comparing, you know, the current AI, you know, separating the hype versus reality and comparing that with RPA. And my version might be a little contradictory. I think with RPA, you still have to actually go through multiple hoops to get RPA into your house. AI is here. Your teams have access to that. I mean, as Steve mentioned, they're already using AI. So I believe, I think there is more opportunity to Joshua's point, he actually mentioned about a contract renewal that actually came up and there was an additional line item. We actually faced a similar situation with one of our providers. And this was around a sourcing platform that we had brought from this particular provider to actually help with our business. We saw that contract negotiation set. You know, we got the tools, we actually built out the major functionality ourselves, and it is already running and going live. So going back, I think this one feels a lot more real and providing value. Now in certain regulated industries, things might be actually moving slowly because ultimately what I am actually seeing is AI is actually speeding up where there is knowledge work, operational work, et cetera. Where it slows down is when it has to actually make decisions and it has to own those consequences of those decisions. That's where really AI is slowing down.
I'm gonna add one thing to what you just said. I think it's building on something that I didn't answer properly. I didn't answer the security piece, which is important. So the number one question I get from executives is, how do I responsibly deploy AI? And I have private equity and I've invested in a cybersecurity firm because this firm, the Oak Trust Group, which was a sponsor of the event, has a different approach to security. And they're working with clients to bridge the gap between what traditional cybersecurity has called for and what new cybersecurity needs, which is a lot more of this governance piece and a lot more framing around. Honestly, I mean, one of the biggest issues is an English-to-English translation between the executive room and either the operating room or the operating committee or the people that are on the ground, that translation of what it is we're trying to achieve. And this firm has an interesting construct called the Oak Trust Group AIQ, which I think I'll explain it to you because it's very simple. It helps with that translation. AI is not AI. If you just say AI to people, then it's too amorphous. There's a cube that they have surrounded by a halo of security or secure by design. The cube talks about what that is, what it is that you're talking about. Is it deterministic? Is it probabilistic in terms of the cognitive architecture? And by the way, generative AI just isn't one thing either. You have constrained generative AI, and then you have unconstrained. Mythos was unconstrained. And what did it do? Okay? That's one axis. The other axis is what agency does it have? If it's decision support and the human in the loop is the king or queen, okay, then that has one level of security and one level of rigor relative to governance implied. If it is completely agentic and there's no human in the loop, that's a whole level of governance that you need to think about differently. And if it's an orchestrator of agents who are doing work without people being able to intervene, that's a whole different level. That's Mythos. So that's the second dimension. The third dimension is what is the scope of its impact? We talk about driving value. Is the value going to be in a domain for a decision maker who is going to make a buy/sell agreement or buy/sell decision on a PE target? Or is it going to be something that's going to be across the enterprise where, you know, for a healthcare provider, for example, they could get clinical diagnoses wrong because a model was off. And then what about the market level impact? Mythos was a market level impact. So those three dimensions, you know, encapsulated with a security or secure by design concept is what this firm is doing, which I think is the right approach, which is protect the data that's coming in, protect the models so they're not going adrift and not being attacked, protect what the output is and make sure the output doesn't go astray, and protect who gets access. Those three or four dimensions of security are imperative. So I think it goes to your point too where if you're going to see value out of this, you've got to bring secure by design into the conversation and governance.
Great. We are at time, but we only really started the conversation. I'd like to thank Eric. Eric will be out there, CEO, looking at emerging technologies, how he's navigating this with his team, a virtual workforce. Josh, dealing highly regulated market, realizing how important contracts and not losing focus of what you've gone through the past and navigate. Grab Josh. Rahul, thank you, bank. Again, highly regulated, the need for security, governance, purpose about where you're making those investments. Aravind, you're seeing this across key industries. You're seeing this from a service provider's point of view about what are best practices your teams are documenting and the capabilities your teams are building up. And then, of course, Steve, thank you so much. Clock speed, I think, is an important piece. Those S-curves, they're still big, but they're so much faster now. How do you govern and secure as you're progressing through that? I think all those are rich conversations to have with these gents during the lunch break, so I hope you will. And on that note, everybody, please give it up for my panel. And we are going to take a break for lunch.
Thank you.
If you don't have an account, Register here |
With the exception of our Horizons reports, most of our research is available for free on our website. Sign up for a free account and start realizing the power of insights now.
Our premium subscription gives enterprise clients access to our complete library of proprietary research, direct access to our industry analysts, and other benefits.
Contact us at [email protected] for more information on premium access.
If you are looking for help getting in touch with someone from HFS, please click the chat button to the bottom right of your screen to start a conversation with a member of our team.
