Video: The AI Payoff: What Separates High-Performing ITSM Teams | Duration: 1789s | Summary: The AI Payoff: What Separates High-Performing ITSM Teams | Chapters: Welcome and Introductions (4.0809999999999995s), AI Payoff Paradox (85.881s), AI Report Findings (184.98600000000002s), AI Efficiency Gains (276.731s), AI ROI Evaluation (435.38100000000003s), High-Performing Teams Practices (612.1759599999999s), Measuring AI Outcomes (866.996s), AI in Practice (1142.2059000000002s), AI Investment Strategy (1487.431s), Implementation Best Practices (1622.136s), Closing Remarks (1697.6860000000001s)
Transcript for "The AI Payoff: What Separates High-Performing ITSM Teams":
Hi, everyone, and welcome. Thanks for joining us. I'm Jasmine, and today we're gonna be diving into some of the findings from our 2026 state of ITSM report, and more importantly, what those findings mean for IT teams trying to turn AI investment into real operational value. So, I'm joined today by Andre Koonen, CRO at SolarWinds, and, Sean Sebring, global leader of ITSM solutions and, engineering. So, Andre, Sean, thanks for being here. Why don't you each, quickly introduce yourselves? Awesome, Jasmine. Thank you so much, and it's fantastic to be here with everybody. Yes. You know, Andre Kuenen. I run our global go to market practice and, super excited to be here. I'm based in Denver, Colorado. And, and we think about global go to market. Right? We have different solutions. ITSM is a key part of our pillar helping IT operations team to be more impactful and, drive value to the business. So great to be here. And thanks, Jasmine, and great to be here with you, Andre. Sean Sebring, based out of Cork, Ireland, and supporting our ITSM business unit as the global manager for solutions engineering. So excited to go through the state of ITSM findings with you guys for 2026. Awesome. So, let's get into it because this year's research really uncovered a pretty interesting contradiction. So we see an AI payoff paradox here. And, Andre, what do you think about some of these stats? That's a lesson. Right? Super insightful and, you know, what I wanna say first. Right? The survey we did is to all kinds of ITSM customers, not just ours, but also some of our, other vendors. So those are global stats. So on one side, you see, right, 84%, right, say, hey. AI actually has met or exceeded our expectations. But on the other side, they say, hey. It's actually not the reducing work hold necessary. Right? And it's not really that surprising. When you think about it, with AI, not just for ITSM, also in other domains of IT, we're moving from an operator to an orchestrator model. What does that mean? That means, you know, in the previous, previously, right, an operator basically did the work all themselves. Where today, they get augmented with AI. Right? So, basically, AI helps them to be more effective and efficient. However, they still need to validate outputs. Right? And they have still human in the loop very early on, obviously. That might improve over time. It'll be beneficial to improve over time. That is the reason you see this kind of discrepancy, which is maybe surprising but not unexpected. And I think, you know, you will see improvement over time as AI is becoming more autonomous over time. Awesome. And you can really see that shift, being captured, in this year's report. And we're not asking whether AI can help. Right? We're asking what really distinguishes the organizations turning that help into, measurable operational payoff from those still managing all of that overhead around it. So, we'll tackle this in three parts. First, Andre and I will look into a few of the findings, to see what stood out the most and what they mean from a business perspective. And then Sean and I will dig into what those organizations are seeing, a stronger payoff than what they're, doing differently and what that means for ITSM teams in practice. And so finally, we'll leave those, slides behind and show you what some of those ideas can actually look like in an ITSM workflow. So let's establish one thing first. You know, most respondents say that AI is meeting or exceeding their RI expectations and they're reporting that these measurable outcomes are, you you know, you see productivity gains. So, Andre, when you look at these numbers, we're talking about a few hours saved here and there across these individual tasks. So from a business perspective, how meaningful can those gains become when you start applying them across the entire IT organization? Yeah, Jasmine. A great question. And, you know, let me maybe take a step back first. Right? So the first step here is 86% use generative AI in Akia san. Rather and I think it's fair to say, in IT, right, service managers, one of the early adopters of AI. Right? Which is great to see. So early adoption, and then when you double click on it, it's okay. Where it's actually helpful and where it also drive efficiency. Right? We saw across the board, right, that has that happened before more manual are now more automated. And you have, like, you know, between three point three and two point seven hours saved the week for those tasks. And think about it. Right? You have many tasks you need to tackle and you add all those savings up. It's actually quite significant. And let me just give a couple of examples. Right? Detecting and flagging issues. Right? You have you have instance getting reported. Right? Instance that need to be, you know, correlated to a problem ticket, and use AI to help with that, to to the correlation and making sure, right, you have a a a more automated approach of, you know, creating problems and correlating incidents. Right? The other piece is you're helping an end user actually logging a ticket. Right? Where they now, rather than using user interface, can use generative AI, just talk to an agent to explain the problem. And then, you know, that that that agent then creates a trouble ticket for us, on the user's behalf. Right? Then you go on triage and categorization to make sure, hey. Now a problem is identified. How do I make sure I give it to the right, people with the right categorization to make sure I really, you know, reduce mean time to resolution for their problem. And that even if it sounds really simple, a lot of IT organizations are still struggling with that, making sure I find, hey, in which part of the IT stack the problem is in, to give it with confidence to arrive a person. Right? Is it the network issue? Is it database? Is it application? Is it infrastructure? Is it cloud? And so forth. So those are just some great examples here on how on very practical tasks, right, AI is helping drive productivity. And when you go back to my previous comment about going from an operator to an orchestrator, right, think about the orchestrator is using AI to be more efficient doing certain tasks. Right? And then, typically, what the operator has to do, they have to validate the data output, but, you know, the output itself is generated by a digital teammate, so to speak. Awesome. Thanks for sharing that, Andre. And that's actually what makes the next finding so interesting because, despite those very real productivity gains, that additional capacity isn't necessarily showing up in in ways that we might expect. So where do we think this time is going? Andre, this seems to change how we should think about ROI. If AI saves three hours and one part of the workflow but introduces additional work somewhere else, how should, leaders be evaluating whether they're actually getting the payoff that they expect? Yeah. No. Great question, Jasmine. I think, you know, the three stats we see here give some of those answers. And think about it, you know, it's like it's like, you know, the journey of getting value of AI. First of all, you need to set up the AI capabilities and also do the respective integrations to make sure you can actually take full benefit of it. Right? So that's typically the people that own the ITSM platform need to set it up so you can take full leverage of it. And integration, right, is one simple word, but integrations can be complex, right, as we all know. The second piece is once it's set up, right, then while AI gives you a lot of feedback and suggestions, you're still in the stage where you need to have a human in the loop. Right? They make sure reviewing and validating the output the the outputs. If you say, okay. That recommendation, I can actually stand behind and, and approve it and then have AI, you know, help with the automation. And we believe that that 4047% over time will go down. Right? Think about it. AI gives you every time a correct suggestion for a particular use case, then you may say, hey. Let me give AI more autonomy, right, where we have less human in the loop. So that's kind of topic number two here with the 47%. And the last one, right, you constantly wanna train and find through the model to make sure AI gives you, you know, more and more accurate results. Right? That is that is kind of on the back end, something that's happening as well. Then, you know, garbage in garbage out a little bit as well. So, you know, making sure the input in AI is solid and then also the model you you selected is skewed as well. So those are some of the main kind of areas where time is going today, and we believe, right, that time will reduce over time, and you get even more benefits of AI. Now what's interesting is, right, only 7% said AI cost matches expectations. And in many ways, you know, we also think that this might be a little bit of a reality and perception gap because we just saw before. Right? How many hours you save on different tasks? Okay. You need to spend a little bit of time of, you know, reviewing the outputs, etcetera, and so forth. So I know Sean, right, has been in this industry for for for many, many years, and he's super knowledgeable. Sean, any thoughts on on that particular step from your end? I love talking about this because expectations is an ever evolving thing. And especially when it's around AI, expectations are constantly changing. So if we had taken these surveys months before, that number would be different. Weeks after, the number would be different. But it's it's growing. Right? It's, and and you know what? These things here are progressional as well. Right? So if you look at managing integrations, validating, and then fine tuning, those aren't gonna go away. But to your point, Andre, it's going to shift, I think, towards something that's more expected, and the perception will also change with it. Right? Imagine I'm running my own business, and I'm by myself as a small business, and I'm onboarding my first employee. It's actually gonna be more work for me for a while because I'm not just saying you're gonna start doing all of these things. I have to integrate you into my business. I have to validate you're actually doing what you're supposed to do, and I eventually wanna just fine tune that you're now becoming an excellent employee and my workload's changing. But what is that change in workload? So I think those expectations are expected to change as we keep things going. Thank you, Sean. That was really great insight, and we're gonna hand it over to the next slide. So, Sean, do you wanna talk about, what high performing t performing teams do differently? Yeah. So I wanna take a chance to leverage my certification field. I just finished transitioning my manage profession managing professional. I am a managing professional in ITIL five now is where I'm going with this. And so the reason I wanna leverage this in is because there's a couple guiding principles. And the seven guiding principles didn't change, by the way, for those who are familiar with, ITIL four and now ITIL five. And the first one is very close to number one on this list here, which is start where you are. Right? And so what can high performing teams do? Start where you are. What do we have today? What's going on? Right? Imagine you're thinking about introducing AI or you have just implemented it and you wanna know how to make sure it's most successfully implemented so you get the highest ROI. What do you have today? What's working well? What isn't? Right? Andre mentioned something that'll probably come up one or two more times at least, which is garbage in garbage out. So know what data you have because if you ask AI to look at your data and give you an answer, if all of the answers that you've given it to look at in the past are no good, it's gonna give you no good information. Right? So you need to make sure you have that strong data foundation. And then something else, again, another key theme that we'll see throughout is making sure that the systems, right, wherever the AI is plugging into or is a part of, integrate as seamlessly as possible. So the ideas behind this is, make sure we we know what we have and we're we're optimizing what we have before we start automating it. Right? So a couple things in practice and, of course, we're gonna look at some of this in in in action later on is proper categorization. Right? Before you expect the system to be able to categorize it for you, right, before AI can do that, you have to give it good examples of what is the proper categorization for a ticket like this look like. Right? And then something I wanna hit on towards the bottom is the measurements. So what what are successful high performing teams doing? They're measuring outcomes, not just activities. And this was really apparent based on what Andre talked about earlier because there is that contradiction of 84% were happy, but then we saw a large percent said I'm doing more work. So which outcomes are we measuring versus activities? Right? The activities would be time saved here, here, here, and here. We can't just measure time saved here, here, here, and here. We have to measure the outcome overall of is the quality of work we're doing better, right, not just that simple productivity. Is the satisfaction increasing? Are we building towards and scaling in the right direction for more meaningful use of our AI? Awesome, Sean. Thanks for sharing that. So we know that good AI doesn't really necessarily replace good ITSM fundamentals. Right? It really makes those fundamentals even more important. So just wanna take you guys to this slide here. There's one more finding I wanna spend more time on because I think it's one of the most important and interesting in the report. Every organization we surveyed measures AI impact in some form, but only 21% describe their approach as outcome or experience oriented. So most primarily measuring activity or projectivity, things like ticket volume, respond times, ticket resolved, or time to resolution. And then we found this. Organizations using activity focused measurement were about 2.4 times more likely to report increased workload. So the findings points to an important relationship, but it does not establish causation. So, Sean, why do you think that distinction matters? This is probably one of my favorite things to talk about, and it'll it's been echoed and will echo here again, throughout the findings from the state of ITSM. Before I do, two quick themes from the transition from ITIL four to ITIL version five. The two major things that came as a difference were a focus on experience management, which is really what we'll hone in on here. And then funny enough, go figure, AI. Right? So AI obvious, there needed to be focus added into our foundational practice. But for the experience management piece, that's exactly what we're talking about here. 2.4 x. If you focus on just those same metrics we we used to measure against without AI, your mean time to resolution, your, time to resolve, first time to response. Right? If we focus on those, we did see ROI. Right? We we saw them. There was hours gained. But if we're only focusing on those and then we see that the workload is increased, right, that's that's where we're gonna see some of this. And I I think if we see to the left there, there's that 21%. Only 21% of them are measuring the outcomes or experiences. Right? An experience I like to dive into is I am the technician. I had fifteen hours on average saved based on adding this AI, but my workload feels like it's more. Why? Right? Asking the question why. They're called signals in experience management now. So what are some of the signals from the technicians, from the agents, from the folks doing the work? And what are some signals, now that we have AI involved, that we should be measuring, to give us what are those experiences like. Right? What are those outcomes? Not just the same metrics, but now with AI applied. So this one's really fun for me because I think it was natural and it makes the most sense from an ROI perspective to measure things that we're used to. To see, did we get a gain? Now that we've done that and we've proven that. Right? This state of ITSM report shows that very clearly. The adoption prop process is is well underway and folks are are very comfortable with that now. Now let's shift and start doing more of the outcome and experience focus. Yeah. And, Sean, let me quickly chime in here. Right? I think, the way I think about this as well, as we saw before, it kinda all starts with activity, rather than how is that activity more efficient. But when I think about outcomes, might only 21% measure it. Right? In the end of the day, you would think, what are the two things a lot of ITS ten teams are measured on? Mean time to identification and mean time to resolution. Right? And you should see both those stats improve, right, with the help of AI. Yeah. Exactly. So, before we move on, we'd love to hear from you guys. Thinking about your own ITSM environment, what do you think the biggest challenge your team faces in, getting more value from AI? So we're gonna launch a quick poll for you guys. K. So so, Sean, let's put all of this together. If you took the findings we've discussed and translated them into, a few principles for how AI should actually work inside an ITSM environment. What would good look like? So what's great is the first three things, if you look at the top and then the bottom left, those first three things are all excellent examples of what I'm about to speak on which is, okay, well, let's embed the AI in the work. And the best way to do that is if it is integrated seamlessly. So we're gonna reduce the complexity of how I interact with my AI while doing my work. And then if they're all in the same system, now I know it's built on trusted context. It's built on the data that's in the system where I need the AI to do the work. Right? If all three of those things are happening in one place, that's ideal. Right? And that's what good can look like. And then the human part. Right? Keeping the human in that loop there. Right? We're supposed to stay involved. Right? We're supposed to stay there, fine tune, validate. We can't have implicit trust. We can trust but verify, as one of my former leaders used to say. Right? So we need to trust but verify. So this is a great segue for us because I think looking at this kind of stuff in practice just to see it in that kind of all inclusive view and how I, as a technician or, someone who's using AI, can stay in control. Yeah. And, hey, Sean. You know, complexity for me is always an interesting topic if you don't mind me chiming in. But when you think about it, you know, what happened in the last three decades with IT, you know, the environments, the complexity has only increased. Right? And when you think about complexities and exponential functions of how many things need to talk to each other. Right? And that's the reason, right, it gets more and more complex. And the challenge that humans had is, how do you digest all the data? But now with AI, with immense power of digesting data and providing suggestions, we have now a different way of solving that complexity issue as well. Alright. So what does this look like in practice? Let me set a stage for you guys real quick. So I'm a new employee. I'm a service desk technician, and I'm responding to an end user. So for me, as a new as a new guy on the IT team, I wanna use AI to make my job easier. And as the IT managers, I wanna trust that the quality, my new employee is getting is at the highest it can be. Right? So one of the best ways is to leverage AI that has that good data foundation we've talked about. So, again, just setting the stage. I've assigned this to myself. I've got a ticket from a a requester here, and he's having a simple application issue, it looks like. Now we know one of the first key metrics that we wanna keep an eye on is responding to that end user. So here's a great first way that I can leverage generative AI. So, again, I'm a new guy on the team, but my AI has gone through ticket history, and it's found a bunch of things that we've done to answer or solve this kind of problem in the past. So this helps speed up my ability to then write a response to my end user. So, hey, you know, I see this is what's going on. Why don't you try these things? Right? And so that speeds that up. I can read through, validate it, and then post this out. Gets me started. So very very nice. Alright. What's another way? So while I've sent that to my end user, I'm the technician. Let's see what it's going to suggest to me here. Right. So this is the system kinda doing the same thing but just talking to me now. Hey technician, I have some steps that you might wanna try. Walk through these things with your end user. Right? We have some additional troubleshooting steps. And again, what's great is it is also my source. So I've got different tickets where it's pulling the information from, problems it's pulling info from, solutions or our knowledge base where it's pulling info from. Right? And so it's giving me the sources, and they're all from in the same system. And this is what I was talking about earlier when I want to have those three things all happening in the same place. And to go back one more time. Right? We see some tuning options here. Right? Validating this was really strong response, not good. Those kind of things keep that human in the loop, and help us make sure we're getting the most out of our AI. You can also see up here that the system's giving us the ability to generate a problem. So if it's saying, hey. I see three other things. This is that time that's being saved. It looked and saw there's other problems out there around the same thing. I shouldn't say problem because we're gonna generate an, according to ITIL problem record. I see three other incidents about the same thing. Do you want to generate a problem? Right? So if I click this, again, we will talk about time saved. Right? It's gonna generate this for me based on the context. So it thought for a minute. Right? And it came up with some context. It's, titled and described. It even went down here and it attached similar things that it found and and thought might be part of that problem. So coming back in here, right, just a couple more things. You can summarize your your ticket. You can also see that we have suggestions to other parts of the platform. So as a technician, this is giving me tools on tools on tools to make me more efficient. My data is trusted because it's in the right system. I know where it's coming from, and it even lets me see and validate those things. So these are just some resources at the fingertips for me as a technician. Right? How can we put them into practice? How does it feel to trust them? How does it feel to kind of validate and fine tune them? So just some quick examples for us there, Jasmine and Andre. Alright. Thanks so much for the demo, Sean. That was awesome. So, one quick takeaway from you. What can IT teams do today to get more of their AI use? So I am gonna hone back into something I said earlier, and that is start where you are. And I'm gonna add one more of the ITIL seven guiding principles, and that's optimize and automate. And I vaguely mentioned that last time too. So what do you have today? Make sure your AI or your data foundation rather is ready. Right? So having that ready data foundation for AI to consume and then help you with is gonna be essential. So the reason the next guiding principle comes into play is you wanna optimize and automate. It's in that order on purpose. You don't wanna automate something that hasn't been optimized because if it's garbage in garbage out and all you have garbage, you're only going to get more garbage more quickly. So you really have to focus on optimize first then automate. So what do I have? How do I make it in its best form and then give it to AI? Right? So that's what I wanna, again, just hone in on for folks who are looking to get the most out of what they have right now. Awesome, Sean. That's super good advice for how people can leverage the their AI use. So then I'm gonna hand it back to Andre to give the final word. So, Andre, from a leadership perspective, what's one question executives should be asking about their AI investments today? Yeah, Jasmine. You know, the way I think about this, right, it's maybe not just one question, but what do you need to do? What actions should we take, you know, with your team to get maximum value of AI? And three things come to mind. But one is invest in training. I mean, the thing invest in training is, first of all, as we heard, the human in the loop needs to interpreting the AI insights. Right? How can I do this effectively? And the other one is and we actually have, you know, AI capabilities to do that in ITSN. You create AI driven workflows. Right? You actually create AI helps you to create automation, and, obviously, people need to know how to do that. Right? And validate it and put it in production. The second piece is right, companies are concerned at times about governance. Right? I mean, you think about governance as two things. A, AI policies, guardrails, transparency of AI. What is the architecture, right, that the vendor brings and how can they actually, you know, basically, address any governance concerns? I think that's a big one. And then the other one we mentioned is the human in the loop. Right? To have a clear policy. Hey. Listen. On those use cases, we run a 100% human in the loop, and then over time, give AI more and more autonomy as it shows good efficacy and accuracy in the results. And the last thing which goes kinda back to what Sean was saying is focus on data. Right? You know, AI is only as good as quality of data. Right? That's number one. And number two is, right, integration complexity. Right? The more you siloed, the more silos you have, you know, the harder it is to kinda bring it all together in AI. So that's the last thing I would say there. So, you know, we have an exciting time here where we see a lot of benefits already of AI, more to come and more to capture, and those are the three steps we are we're suggesting organizations will tell. Awesome, Andre. Thank you so much for sharing your perspective. And I think that's the perfect place to end. If there's one thing I hope you can take away today, it's that the twenty twenty six data ITSM isn't really a story about whether AI works. The data tells us it really does. Right? But it's about what happens next. So how do you turn those productivity gains into meaningful capacity? How do you prevent AI overhead from absorbing the value you're creating? And how do you connect ITSM performance to outcomes the rest of the business actually cares about? So organizations pulling ahead aren't necessarily the ones buying the most AI or spending the most. Right? They're being more intentional about where they use it, reducing the friction around it, and ultimately strengthening that foundation underneath it and measuring the outcomes that really matter. And, ultimately, just bringing, all the people along with it. So, Andre, Sean, thank you both for joining me, and thank you everyone for watching. You can scan that QR code that you see there to download the full twenty twenty six date of ITSM report, call from AI potential to real payoff to explore the rest of the findings. Thank you very much. Thanks, everybody. Excellent. Thank you.