Name: Hardy Kuhn, SAP
Uploaded: 2025-04-14T17:00:00.000Z
Duration: 20 min 43 s

News
More from From Vision to Reality: How AI Is Transforming Work Now

Hardy Kuhn

Information Technology Executive

SAP

Trust is the new algorithm — and SAP’s laying the foundation with clean data

To build trustworthy AI, the data has to be just as smart as the algorithm.That’s why SAP SE is doubling down on quality control — not just to reduce bias or sanitize inputs, but to ensure artificial intelligence outputs are genuinely useful, accurate and meaningful. From text generation to service recommendations, it’s not enough for AI to simply “work” — it has to work right. And getting there requires a human-in-the-loop approach, according to Hardy Kuhn (pictured), head of service management solutions at SAP.“It’s not just an automation, and it’s not just understanding language, but it’s really then conforming and creating text that is meaningful,” he said

play_circle_outline SAP's culture of innovation and approach to AI use cases

play_circle_outline Adoption challenges and change management for AI implementation

play_circle_outline Results and benefits of AI use cases in ITSM and CSM sectors

play_circle_outline Exploration of agentic AI and future hopes for AI impact

38:35

25:35

Info
Transcript

Hardy Kuhn, SAP

Hardy Kuhn

Information Technology Executive SAP

Join theCUBE's Savannah Peterson as she talks with Hardy Kuhn, head of service management solutions at SAP, to discuss how SAP is transforming service management with the power of AI. Kuhn shares firsthand insights into the company's AI journey, from streamlining customer interactions to boosting support efficiency through automation.

This exclusive conversation explores how SAP, in collaboration with ServiceNow and Accenture, is building innovative AI solutions that balance technology with thoughtful change management. Kuhn highlights the essential ro... Read more

Hardy Kuhn, SAP

Savannah Peterson

>> Hello and welcome back to our Palo Alto Studios here in sunny California. My name is Savannah Peterson. Delighted to be bringing you a special segment from our super awesome series with ServiceNow and Accenture talking about how AI is transforming work. There's a lot of conversations about how AI is going to change the way we work, questions about whether or not it's going to change our job, but the reality is much more complex and cool than that. Here to tell us a lot of this story is a special guest coming to us from Germany today, Hardy Kuhn from SAP. Hardy, thank you so much for taking the time to hang out with us.

Hardy Kuhn

>> Thanks for having me.

Savannah Peterson

>> So when I was doing my homework on you and talking to our mutual friends at ServiceNow, one of the things that struck me was how innovative everyone raves about SAP being. Talk to me a little bit about the culture of innovation at SAP and how that is feeling right now in this AI hype curve era.

Hardy Kuhn

>> That's an interesting one. I think the DNA of innovation was in our company to begin with. We're a tech company, we're in the software industry, so it's natural that this has taken evolutionary steps from early days to today, and we're really excited about the AI use cases and capabilities that are coming up.

Savannah Peterson

>> Let's dive right into those. What are some of those use cases?

Hardy Kuhn

>> Well, I'm responsible for the infrastructure for our IT support and support agents supporting our customers. So that's where the collaboration with ServiceNow is going on. I think some of these use cases are tailored at making our agents smarter, faster, better in serving our customers. Coming from the customer experience side of the house, obviously this is the goal that we're pursuing. The use cases are manifold. In the beginning, very small and easy ones, and they get more complex as we go on.

Savannah Peterson

>> Can you give me some examples of their use cases? Because I think you're someone who's seeing this implemented at scale, you're seeing the benefits and everyone's talking about the potential of AI a lot, but to actually have achieved results is a little harder conversation to find. So what were some of the use cases that you identified right off the bat and how were you able to choose? Because there are so many different applications for AI.

Hardy Kuhn

>> Absolutely. I think our business of serving customers in cases or in service management is very tech-savvy. It goes back and forth between our customers and our agents, and there's a lot of writing, a lot of text, understanding each other, et cetera. So I think the very first case that we started off with, AI use case was a case summary. When we have a handoff of agents so that they don't have to scroll through pages and pages of going back and forth, but we put the text, summarize it and put it in a short format that is very structured and is helping our agents then to be faster in grasping the situation at hand.

Savannah Peterson

>> Yeah, when you're troubleshooting something you want to have the most information available, I mean this whole purpose of a knowledge graph-

Hardy Kuhn

>> Absolutely.

Savannah Peterson

>> And it makes a lot of sense. It relieves that stress and anxiety in that moment too, that you're able to deliver the right information in real time to enhance the experience on both sides of that. What's the reception been like for the agents on your team?

Hardy Kuhn

>> I think that's a more interesting question because AI is not embraced by everyone right off the bat. There's a lot of anxiety, a lot of fear, a lot of mistrust, but also a lot of excitement about these things. So we've seen this is a change management just as much as a technology challenge. We've seen a huge uptake in the very beginning of these AI use cases, but we did a lot to begin with. So we formed a team. We had some little use cases upfront as a playground, so to speak, to get the team on board and familiarized with the technology and capabilities. And then when we came to the business, there was in the very beginning a very long testing phase. So it wasn't right out of the start and went live. So we had a very long, maybe overly long testing phase of these first AI use cases just to make sure they don't fantasize, that they don't tell the wrong things or things like that. And then I think that's when the first go live happened, when everyone was happy. And ever since we've seen a huge uptake in usage of these first AI use cases.

Savannah Peterson

>> I'm curious because you say very long but we've been living in this innovation cycle for not that long. Well, maybe you've been doing it for a long time. How long was that testing cycle?

Hardy Kuhn

>> Actually about two and a half months. Compared to how we built this one use case, I think it was equally long. So we tested just as long as we did. It did take us to build it and I think that's what I would consider long in this space.

Savannah Peterson

>> I think it's impressive that you're considering sub-90 days long, especially at the scale that you're at. So that says something about the level of innovation that's going on over there for sure. Thank you for being candid about the agent side of things. I think adoption of any new technology, software tool, whatever it is, requires a bit of prep across an organization in order to achieve that value quicker and with the least headaches and with the most enthusiasm. How have you prepped the rest of the organization for this transformation?

Hardy Kuhn

>> Well, I mean prepping starts with common knowledge about what's out there and I think that we had a lot of headwind of lots of talk about ChatGPT and its capabilities two years ago. So everything from there was a major rollout of what we can do, what we will do and how we go about it. I think there was also a lot of communication about data privacy, about security, about how we treat data in a certain respect that they're not misused, to take away that anxiety of should we, could we do this? And then I think it was what is the benefit if we do so? We did a lot of value calculation on this part from the get-go. So what we had use cases and we articulated what the objective was. We then started measuring adoption and business KPIs for each of these use cases to see if we really get to where we want it to be. And I think that all needs to be communicated transparently, and then I think you have your rollout strategy.

Savannah Peterson

>> I think you just nailed it. I'm sure anyone listening right now is like, okay, note to self as we roll this out at scale. That was excellently stated, Hardy. I'm curious, were there any surprises? It sounds like you were very vigilant and diligent in the rollout here. Was there anything that surprised you about this process?

Hardy Kuhn

>> I think a lot of stuff, I mean the 90 days that we were talking before was just one use case, but we are on the road for almost two years now. So we were one of the very early customers to embrace this one and probably one of the first AI lighthouse customers in general. So we were out there before the technology was mature. So I think for us personally, it was not a surprise, but this was more a co-innovation journey. It was really shaping the technology with the use cases in mind. So our partners here helped us greatly really coming up with the first version where we could toy around with prompting, with using data, with then looking at the different services that we could and could not yet do. What is easy to do and what isn't? So we staggered our use cases in complexity. Everyone wants the holy grail of automatic answer without intervention of humans, but I mean that would be in most cases the most efficient part, but nevertheless, I think you have to start to crawl before you can run, and I think that's what we started to do. So not surprising, it was a journey, setting up the team, setting up the technology, then toying around with it, then really getting serious about the first use cases. And then finding out not everyone is excited about it, there is a lot of anxiety, like I said before. It is also then change management to deal with that or adapting the use cases or the technology to that. I think that was pretty much what we discovered in the year and a half that we're underway.

Savannah Peterson

>> Talk to me a little bit how that partnership came together and what it's meant for you to have these partners on the co-innovation journey with ServiceNow and Accenture.

Hardy Kuhn

>> Well, luckily we were already underway with the two partners for quite a while, so this was not a new relationship that we established. I think our ServiceNow journey is now going on for five years. We're using it in the ITSM sector of it, so really service management, case management and interaction with customers, but also internal use cases. We've implemented and pulled in a lot of stuff in-house to the ServiceNow platform as a backbone of this, and then we had a platform to innovate on top. We are a technology vendor, so obviously we bring AI capabilities and AI technology with Joule as the chatbot and with our business transformation platform that has now a GenAI capability to the table and we have to combine both of these technologies for the better of the use case. And then Accenture was our implementation partner from this journey, and that's where the journey started on AI and GenAI especially.

Savannah Peterson

>> I'm curious if there were any... You talk about the adoption from the agent side and the evangelism across the organization for the adoption of these new programs. Have you gotten to see any cool wow moments where someone on the team really was able to understand or see how much AI was going to enhance their job or make their day with more purpose?

Hardy Kuhn

>> Absolutely. I think we were all surprised how good these large language model are coping in dealing with languages. And then as we tapped into our knowledge base and we enhanced these capabilities with our own data, I think the answers even became more sensible, more meaningful, et cetera. I think that was really surprising. It's not just an automation and it's not just understanding language, but it's really then conforming, creating text that is meaningful. We still have though, to be on the safe side, the four-eye principle or two-eye principle I should say. So we have the artificial agent that makes a proposal and we still have a lot of these use cases serve our agent, our human agent as a proposal, as a human in the loop to make sure that we have a quality control before it really goes out. So I think we're still in the journey, as we all are right?

Savannah Peterson

>> Oh, yes, we are still in the journey. We are very early in that journey, I would argue overall and we're all learning together, which is actually kind of the fun part. I think the collaboration and co-innovation that you were mentioning earlier is one of the coolest parts about this particular technological revolution, and I know that it gets me excited. Speaking of our... Actually, well, before I shift gears, and since you have been working at this now for a while relative to some other companies, can you share any results from these efforts and this rollout and the lighthouse project?

Hardy Kuhn

>> Yes. We have a lot of use cases in the ITSM and CSM sector already. I think five are business live. With these, we've seen an adoption that is really a hockey stick. So really, really great adoption internally. Lots of agents like it because we collect the feedback as well. For each one of these, we have a thumbs up and thumbs down mechanism. You can even type back, did I like the answer? Et cetera, so that we have a little bit of a qualitative measurement whether agents see a value or perceive a value. And in addition then, we started measuring calculated times of savings. And these might seems small in the beginning when you look at an individual case of let's say a summary saving nine minutes per agent or something. Rather than reading through the whole thing, you have to read through a smaller, more structured segment. But you multiply that by the amount of cases and the amount of agents that we have. I think it already becomes a huge number. So we can prove the value in these cases. The same is true, we have a use case live where we create knowledge articles out of solved cases and the time saved of just copy and pasting certain fields into a knowledge article that the agent had to do manually before, is perceived and realized value that you can actually really measure. Is it perfect? Is it a full knowledge article that needs no human intervention? No. But we save 60% of the time, we've come a long way.

Savannah Peterson

>> And 60% of that repetitive time, if I can learn what I need to learn or share what I can learn faster-

Hardy Kuhn

>> Exactly.

Savannah Peterson

>> ... there's no loss there. It's only a gain, and think of all the other things can use their brains for with that nine minutes they just gained back every single time. I think it's really compelling. It does make a difference. We're talking a lot about agents here and it would be impossible to be talking about AI in 2025 and not talk about agentic ai. How are you and the SAP team approaching that and implementing new strategies and rollouts both internally as well as across your community?

Hardy Kuhn

>> I think that's the next evolutionary step. We've coped with simple use cases. We chained some of these use cases. So it started with easy prompts, then they became more complex prompts, then they became prompts that were using extended knowledge in the background. Then we programmatically chain some of these. And now I think the next step is obviously agentic AI. Where we have a use case in mind that is more complex and where agents then autonomously need to decide on certain outcome what the next step is, and I think that's where we're starting to invent and build up these agentic AI use cases as well. We're still in the beginning of that early stage, so I cannot really say we've mastered it, but it's the next evolutionary step.

Savannah Peterson

>> Yes, it absolutely is, and I think it'll bring a lot of things to scale and we'll have a lot of little helpers doing some of the tasks that we don't enjoy doing as much throughout the course of our day. This has been an awesome conversation. I'm curious, Hardy, taking off... Well, you don't have to take off your SAP hat if you don't want to, but there's a lot of conversations both hype-filled, theme-filled, chaos-filled and inspiration-filled around AI right now. What do you personally hope our AI future brings to your family, to the world, to the people watching this, whatever that might be?

Hardy Kuhn

>> Well, I know there's a lot of anxiety about it. When I talk to my parents about it, they expect the worst from it. There's a lot of people having the anxiety that they lose their job as it is today, and I think that's the fear that we have to deal with. I think personally at the moment I see that in my calendar, in my daily task, there's a lot of repetitive stuff that I would love to spin off to an agent and get it done, and I could concentrate on more value add or more creative or more thoughtful processes in the end. So I consider this as the next tool of automation and next helper, so to speak. And will it transform jobs? Absolutely. But I hope for the better, and I think that's where my real responsibility to all of us in this industry and to all of us who create and use and shape these AI moments, you have to really think about data, data privacy, how you use it, responsibility of these things, how they're used and what they do. And I think then we can shape a fantastic future where jobs become enriched and you get help rather than have angst about it.

Savannah Peterson

>> That was beautifully said, Hardy. Seriously, you nailed it. There is a real sense of duty and responsibility for all of us. But like you're saying, I mean, we do hope that the efforts of all of us are a part of that good change and that good future progress. Last question for you, just because you are in Germany and I think you're our only guest on the show for this series in Germany, how's the general conversation around AI in Germany? It seems to be all anyone's talking about in the United States right now.

Hardy Kuhn

>> Same as here. I think I'm now caught in the bubble of being in the biggest tech company in Germany, so naturally the conversation is a little bit different. I think when we go outside of this, and I can just speak about my personal and about the media and stuff like that that I perceive, there is a lot about respect for this technology, for the change that it will bring maybe to society and to learning and to jobs, to functions. I think a lot of people are scared, and I think that expresses extremely in Europe and in Germany where we start regulating the use of this technology before we even understood it or have scaled it or put it to use. That's my personal opinion on it. On the other hand, as I said before, that social responsibility that we all have, I think is an expression of that. So I would be excited if we could put this to productive and good use.

Savannah Peterson

>> Yes, hopefully we can all continue to stay excited and keep having these exciting conversations about the innovation and progress you're making over there at SAP. Hardy, thank you so much for taking time this evening to come and educate us and impart your wisdom. This has been fantastic.

Hardy Kuhn

>> Thank you very much.

Savannah Peterson

>> And thank all of you for tuning in wherever you might be. Hardy's in Germany. I'm here in Palo Alto, California. Delighted to be bringing you our special coverage from Accenture and ServiceNow, discussing how AI is transforming work. My name's Savannah Peterson. Thanks for tuning in.