Interactive session · 11:25 AM · Wednesday, May 13, 2026
Phil and I are going to just ask questions. So look, I think we've been talking about this. There's a lot of hype. There's this dichotomy that we are seeing where everybody wants AI, but nobody's getting AI. The results, right? All the three, four sessions today have really talked about that. So we, and there's so much data, right? It's like 92% of the enterprises that we reached out to believe in agentic AI. But only 7% have got there in some form or fashion. So we're not even solving for a snacking right now. You know leave aside world hunger.
So what we wanted to do was to do some polls, live polling. And I will roam around and Phil will roam around and ask you questions on why this is happening. So if you can get your phones out and start polling, and the first question is, what is the biggest load of agentic AI hype you've been sold? Is it agentic washing, you know, everybody's AI first, everything is agentic. Every RPA company is now an agentic company. Every workflow company is an agentic company. HFS is an agentic company. AI first, you know, how come everybody's AI first right now? It's just not statistically possible, right? Somebody's got to be AI second. There's got to be one. I think AI second is a much more differentiated positioning. The other thing is, AI will, as I mentioned, solve for world hunger, you know, eliminate toil, debt, suffering that we all face in our jobs, or also that, you know, person X is way ahead of you. You know, you're falling behind, you're a laggard. So let's see if you can poll, and Phil, do you want to find some guinea pig? Here we go. Everything. It's evenly distributed.
Okay. How are we going to get started here? Pick on somebody. Who's voted and has a big opinion on this one? Michael?
Yeah, I think the agentic washing has, you know, everything is AI. It's kind of like everything was CRM and everything was RPA. And everybody, every CEO is asked the AI question. So this agentic AI is solving everything, is going to reduce cost, is going to make us more competitive, is a false lens. You know, we're starting to see the, you know, you just heard the J-curve reference before. But I saw another study that was looking at, are you prepared for the AI winter? So that's, I think, going to be this disillusionment. Why didn't revenue grow? I thought the Accenture comment just now was the biggest. You know, and we had the conversation three years ago, if we're spending $3 trillion, where's the ROI? And exactly what Manish said was, it's cross sales, new sales, and growing revenue. Just focus on that, period, because that's where the value has to come from. That's huge. I think that's the biggest moment we're going to get. That takes away all the washing and everything else we're looking at.
Yeah, fantastic. I've got Ranjan here who's been sitting in the back from Citibank. What do you think?
Yeah, hi. No, thanks, Saurabh. I think going back to what was discussed earlier, I think there's too much of a focus on a functional view, which is why you don't have as many agents in place. I actually voted for agentic washing. And when you have a functional view versus a business view, then you're really getting down to fragmented areas of opportunity. I think you have to look at the business process first, look at revenue generation as part of that, differentiated customer products that will drive more revenue, and then get down to designing that business process that will drive how your whole end-to-end works, the business and in the functions. Otherwise, agentic washing is really, I think it's not as well penetrated as we've been hearing.
CJ always has something to say.
Thank you. So my understanding is that when someone is looking at AI deployment, it has to be distinguished between the structural challenges and the one which is executional. So when we look at the structural challenges, I think it is important to look at as to how is it we can reduce the debt. And I have written about this organizational autophagy. You really need to get some of those things eliminated first when we talk about the structure. Otherwise, if you look at the executional, it is important that we have the right kind of AI, pre-built AIs which are already available. Those can be deployed. So that's my view.
Fantastic. Should we move to the second one, Phil? I think so. All right. So we've talked about the biggest hype. Now let's talk about what's the biggest thing holding you back. And I think, I don't know if Manish actually has read our research, but his five points were so aligned to what we've been saying, Phil, is these four debts: there's a technical debt, there's a data debt, there's a process debt, there's a talent debt, and perhaps there's accountability or leadership debt, which is number five, or maybe it's all of the above. Look, my point is very simple. At some point of time, we have to pay our credit card bills. You know, you can't just keep buying off your credit card. At some point of time, you have to pay that. But what's the opinion of this room? Which of these or something else is really holding us back? Oh my god! What is it? Talent! All of the above. Everything. That's a hard problem to solve. Who's got an idea to solve it? Debbie, how do you solve all of the above?
You need Accenture, no I'm kidding. Sorry, I think it's gonna take a long time because you're not gonna solve all this in one fell swoop and I think that was Manish's point. You gotta start with all the things you're in debt on, you gotta build your talent and you gotta change how the company operates. I mean, there's just no quick answer. And I think that's why this is a good opportunity for those folks in this room that are in the services side and potentially the software side as well.
Can I add something here? I think there's a big company, sort of small company mentality issue going on here. If you work in a big corporate, why should you put your job at risk? Why should you care? You know what I mean? You'll spend a lot of time educating yourself, making sure that you can use these tools and everything. But unless you're given real incentives from your boss and your corporate leadership to do anything different, you're just going to carry on rearranging the deck chairs as you were in the past. Whereas a lot of people are going to smaller, mid-sized companies now. And I'm sure a lot of kids coming out of college are getting job offers from small companies. Because they'll hire three people in marketing. And then they'll grow these companies with an AI mindset. Whereas large companies, you've got to get that more agile, smaller company mindset to be effective.
Phil, what's also interesting is the technical debt sliver is the smallest. We've been talking about technical debt for forever. What do you say?
I wasn't going to say anything, but because this one looks so impossible, I felt compelled to jump in. I think it's so freaking doable. It really is. It's pretty much taking each one of these five challenges one by one and tackling it. In the company that I work at right now, Under Armour, what attracted me there is that they took the data debt. And they started preparing the data for this moment in which we're going to do AI seriously. So when I walked into the room, when I say AI, I mean Gen AI. Gen AI was not a toy yet in the tool set. It was just the focus on data debt. And that foundation was so strong, then I started to play with the other areas. Technical debt. You look at the renewals. What's coming up? What can I eliminate? What can I try to? Can I create a timeline in which I try to moonshot something? And if I fail, we renew. If I don't, we don't renew. And we were able to win one of those challenges of a big, big renewal that we were able to eliminate. And that created the reputation of this nascent team that nobody knew. The process that is what Phil just said. Can we invite people over to be part of a new process instead of this message of replacing people? And if you're a CEO or in the C-suite in this room, can you please calm the down about the firing people for AI? It's not realistic. Can we please stop this nonsense? People that actually understand the difference between machine learning and generative AI know that everything that people are saying out there that's deterministic, it's not really. It's just a wrap of marketing. And you need a human in the loop, you need people doing thinking around us to really create this incredible shift that we're about to go to. And to wrap it up, the talent that, when you create processes that involve people, that empower them with AI, it becomes a cool place to work that attracts new people. And then we get closer to this future that we want to get anyway.
That's fantastic. Should we just wrap up there? But I've got Marisa here. Sure.
So I would say a lot of the points already made I would agree with. I did vote for, it's probably a little bit of all of the above that has to be worked through. It just depends on which one is your bigger debt at the moment and what you prioritize. I do think leadership debt needs to be back up here, though. I know that was part of the conversation because how leaders are going to help lead through this change, position for it, and support is going to be really important because all of these will be factors over the course of the next, this is months, not years, that we have to think about. So I think working through the hype piece from question one to be able to best position your strategy for this question two is going to be really important.
Can I just add? I'd like to add a point, and I was reading the green one. Nobody owns the outcomes when an agent gets wrong, though I mark talent and leadership debt, but I see those two very connected. I don't see a situation ever that you identify a process or a function and there isn't a human leader or a middle manager who still has to own the outcomes. And for me, I think this is going to be the biggest step change that organizations have to go through. Are the leaders going to feel comfortable and what skills they need to have in order to manage the agents and the outcomes? I think that is the, that is going to be the one of the bigger stumbling blocks. Technical debt, process debt, we'll surmount those, but that's the tougher one.
Interesting. Hi. Quick question about, I want to correlate this with the, one of your earlier slides this morning about high maturity versus low maturity organizations. What does your research show in terms of the, let's say very mature organizations? Have they been able to fix all this debt? Have they overcome this debt? Are they doing it? Is it a process? Is it, you know, how does it comport with this?
So I can, I think I'm gonna make it up. But, so we've, this is a research that's not yet published, but actually next week we are doing a round table to sort of start launching it. But only, we reached out to about 2,000 enterprise leaders and asked them on these four debts. We didn't have leadership as a question. Rightly or wrongly, we didn't ask that question. And nine out of 10 of those 2,000 people said it's a big issue. And almost all four debts are equal. So there's only about six to 10% who said, we've solved for those. And I think all of those were at the C-level, or large majority of those 10% who said we've somewhat solved for them, were a C-level sort of respondent, who were saying we've looked at it at an enterprise-wide thing, this is a CEO mandate, this is not an IT project of legacy modernization or moving mainframe to something else, it's an operating model change. You know, the kind of stuff that we've been talking about. So they've really invested in that. And I think one of the first things that we also saw from that research was while talent debt might not be the sort of biggest sliver individually, that sort of trickles to everything else. So if you don't have your workforce, skilling, planning, et cetera, it doesn't give you the immediate ROI. But that was another difference between the 10% and the 90%. So I don't think there is a, to Debbie's point, there is an easy answer to this. But one, being cognizant of the fact that this is an issue and this is what's going to hold you back. And then having a little bit of a mature conversation on this than just making it like an IT project or let's now solve process debt. That's not going to really work. It's just a question, not a comment.
So when you looked at businesses, was there any difference in business to business companies versus business to consumer, where there's that connection with a consumer?
So in that study, the top two industries with the maximum debt was manufacturing, was number one. And the second was healthcare. So I'm trying to relate to your question. I'm just looking at that data. And I think it's those companies that have been just doing the process-driven. You know, even if you look at manufacturing, it's a very process-driven approach. And same with healthcare. It was healthcare providers, actually, which was second. It's always the process. So I think it's less to do with B2B, B2C in my view, Debbie, but more to do with companies that are just going through the motions versus doing... Like it was actually lower for services companies than like banks and who we would think has a lot of this debt on a relative basis was lower.
Related to this, all of this is dependent on individual and teams to perform. Now, one thing that I find missing here is if the rewards and performance metrics of an individual and the team is not measured on outcomes and AI successes, right, and they are still measured using the legacy mechanism of how many people they control, how many projects, et cetera, then there is no inherent motivation to make AI successful. And I think that could be one of the biggest roadblocks. So what would your commentary be around that? And second is around change management. Because the top management does say AI is a priority, but how does the change management need to be done at the middle management and the larger broader teams, especially with global organizations?
Yeah, this was supposed to be a session where I ask questions. It's just turning it around. But I think you're absolutely right. The incentive model is a huge challenge right now. I think the incentive models have been designed to incent people who score the goal, not incent people who pass the ball. And I think that is a fundamental shift that we'll have. Everybody keeps talking about we need to collaborate, we need to collaborate. But why should I collaborate if I'm not incented to do it? If I'm only incented to just win the new deal, and then there's this fighting that, "Oh, it's my deal" and who gets the commission and all of that. So I think most of our models are based, incentive models are based on scoring the goal versus passing the ball. And I think that needs to change. But on that note, let me reverse the cycle again and ask questions. I'll go to Umang. You're with JPMorgan Chase. What are you seeing on this?
All of it. But JP Morgan, right from the top, we talked about CEOs. Jamie Dimon has been very clear. If you look at it purely from a dollar allocation, they are allocating more than $2 billion on a just under $20 billion tech budget. So that's 10% purely on AI. And there's also a comment from him, it's not verbatim, it's, this is not just true for AI, but if you go historically all tech projects, sometimes it's very hard to quantify ROI. And we just got to be very clear about that reality. Because there's an entire big conversation when you look at the whole concept of how much does it produce to actually create a project, once the project, it may be a product as well once it's released, what's the cost of management of that? And then how do you account that through the ROI lens? And it's not the first time we are dealing with these questions. What about 10 years back when we went through the cloud journey? We didn't know what to put on prem, what to put on cloud. Is it secure or not? Do we have talented people to actually make that happen? And then the industry itself evolved. There are hyperscalers like AWS and others. They actually came up with certifications, which became an official way for people to get certified and then they can actually go and implement. We are going through the same journey. And along the way, at that time, process gaps, data gaps were identified and fixed. And we'll probably go through the same process again. And will everything get fixed, especially your legacy products for the last 20, 30, 50 years? Probably not. But would you fix them when vulnerabilities are being found? You know, we all know about the model that came out, was released for a few companies, and a lot of the vulnerabilities were exposed, and companies have been working to fix those. And this essentially is a patching exercise you have to go through as your company, big or small, goes through the whole maturity model, and we're essentially going through that whole cycle. But from a company, and what Jamie has been saying is we need to invest dollars in this because after a while, if you are working with a legacy product system process, the cost to just maintain that, and again, this is my personal opinion based on my background, the cost will be so high that you'll be paying a lot more to maintain that versus just migrating to a solution that actually is much more efficient, is much more cost effective.
No, I completely agree. I think most firms, including us, we estimate technical debt to be $1.5 to $2 trillion within the Global 2000 firms. And so what have we done to solve that? We've created another $1.5 trillion IT services industry. And 70% of that IT services industry essentially maintains that debt. There's ADM, there's IT infra, there's service desk, etc. That's just maintaining the debt, not solving it.
We're obsessed with money. You saw me this morning. America has made a giant bet on AI. Trillions of dollars are being pumped into AI, in the infrastructure and compute. Where's the money going into their education? Professional training. The real issue right here is there's so much excitement around the fact that we've got some technology that does lots of cool stuff. But the real issue is people need to change and learn, and our education system needs investment. Our businesses might need some support. Where's the investment there? Rohan wants to chime in.
I think we're also focused on what's happening here because we all live here, but if you just think about what's happening in China and DeepSeek and how quickly they are innovating, I think part of this race is to just stay. If you listen to Jensen Huang speak, he's very clearly articulating the pace at which the LLMs and the AI models that are developed in China with very little governance are already doing a lot more things. So the question was really about, are we spending to just keep parity, are we spending to actually keep spending? The reality is, if you really re-architect your systems from ground up, it's really easy and fast to do it. But the change management aspect is, that hurdle is still human. No way I can solve that. And I think that's, if you just take all of this and you wrap it into a single thing, it's just the propensity of us as humans and the resistance to change. I mean, who moved my cheese just sort of comes back to my mind when I think about that. Yeah.
Thank you. The nobody owns outcomes, I would leave it at that and take away the second part of it. And if there is accountability at the C-suite level across the board for each one of the other areas, debts and there's true accountability. And that the, yes, you have to have an owner for every single element within the other debts and this comes down to the leadership debt and my friend over here said the imagination debt, the curiosity debt. And then there's also, this goes into the leadership debt to ensure that you have a culture, not only with psychological safety, but intelligent failure that's rewarded.
I think we had one more question, so let's quickly do that. Oh, no, we didn't. Oh, we had. So when do you think we'll solve for these debts? Under six months? And if somebody is saying that, please share that secret with us. Six to 12 months, more than a year, no idea. Let's see. No idea. There are some optimistic people, 6 to 12 months. But maybe, Rita, if I can ask you, there's a lot of conversation around C-suites, et cetera, here. You place people on boards. So let me elevate this. Is it a C-suite problem or is it a board problem?
I think it's both. Because right now we have a lot of people on boards who really don't understand AI, they don't understand the opportunities, and they can't imagine what could be done in focusing on revenue. I think that was really telling to hear more that the focus should be on revenue, and we're thinking about that too. So how can we use AI, not for internal processes, but more for driving revenue and opening up new opportunities to work with companies on board searches?
So. Okay. So now we've got boards involved as well.
Yeah, you know, there were a lot of interesting questions asked. And Phil, I think you asked about who owns it. Is it bottom-up employees or is it top-down? You know, when you're dealing with tsunamis, you can't expect individuals to solve the problem. Okay, it's not going to happen. Because first of all, you're asking them to imagine a world where they'll be shooting themselves in the foot. It's not going to happen, right? Or shooting themselves in the head worse. So my take here is somewhat different in the sense that we went through a similar upheaval when cloud happened. And people have forgotten. I remember talking to CIOs who told me, performance is not going to be good. Security, I doubt, will be good. Both of those, by the way, are totally wrong. Never proved to be true. What cloud changed was it virtualized everything. And it took away constraints and barriers. And I think that's exactly what's happening here. Agentic AI is taking away the barriers of a physical organization. Somebody asked who owns if agent does something wrong. There's always going to be some manager somewhere who owns the agent. OK, it can't just be agent did something, what can I do? So I think we need to reimagine an organization which it doesn't have the traditional constraints of physical beings. It's all virtualization. To me, agentic AI is about virtualization of work. Just like cloud was virtualizing everything, and people thought it was about virtualizing hardware and networks, it was much bigger than that. It changed how technology value gets delivered. And I think the same thing is true of agentic AI. So I want to really say it's about virtualization. It's going to virtualize your organization. Organizations are not going to disappear, but the constraints that organizations bring, every time you transform, changing an organization is a nightmare. You won't have as much of a nightmare with agentic AI being embraced. OK? That's the point I want to make.
Fantastic. So on that somewhat positive note, I think we should move on to the next thing. But I just wanted to say we now have artificial intelligence, we've got augmented intelligence, and now we've got virtual intelligence as well.
I just wanted to add to that. I'm not upset anymore, guys. I'm calmed down now. I love that. And I love what was said before about shifts to transformation. When you look at a letter A for under six months, this is really about if you're really committed, it can be halfway. If you commit it, then the first step is to look inward. And sorry, Accenture and all the amazing vendors that we have and partners that help us transform and grow, it starts by inside, but inward. You have to invest in that shift happening from inside, empowering the teams, creating a paradox of a different kind of management. So something that we've been playing with, can we empower single contributors to be elevated to managers of AI systems? And allow them to get the tools instead of feeling replaced when they build these tools with us, they have this knowledge that no consultant has. They understand the systems in the company like nobody else. Can they build a system that elevates them and has a role defined for them? And then above that, last thing I would add, when you talk about security, you're right, cloud, this is reminiscent of what we went through with cloud, but cloud only became serious when we created governance around it. And when we invested in the team, we changed the names of departments, we created new roles. How much time is it going to take you to start that now?
So I think we wrap up now. Thank you. Thank you, Liz. Thanks, everyone.
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.
