and the next day. Good morning, everyone, and welcome to our morning webinar that we're calling Leading with AI. We are super, super excited to share with you some of all of the experiences and insights that we have gathered these last 10 months, couple of years working with AI and putting it in a context of leading with AI. But just to get us started, the brief introduction of us. So my name is Marie. I work in Implement and our data and analytics team have been working with machine learning and artificial intelligence for the past five years. And in the last 10 months, it has been primarily around generative AI. So super, super excited to be here today with Marie. Yeah, with Marie. And my name is also Marie. And I've been working with organizational change for the last 12 years, primarily with the IT implementations. And for the last two years, I've been engaged and driving generative AI and AI projects in general. And I'm super excited to be here. And I'm super excited to be able to share, first of all, the first learnings that we have experienced by working with those projects, but also highlighting some of the challenges that we have seen by engaging with the organizations. But we're also super curious to hear who's out there. So just take a minute to write in the chat, maybe your name, your company, where you're joining from, where we're excited to see if we're reaching outside of the of the Danish borders here. So let's see. We have Hummelbæk. Welcome to Jacob. We have some people joining here from the Nordiskfronten. Amazing. DK, Aarhus. Let's see if we can find anyone that is a bit outside here. Hey, we had one from the UK. Nice. All right. And Oslo. Amazing. Yeah, keep it coming. It's really exciting to see who's out there. And who's joining us this morning. But let's just jump into it. Let's jump into it. So first of all, just to highlight the purpose of today. In the invitation, it's also so state set, but just to ensure that everybody is on the same page. Today is not about how a leader can use AI. Today is about how a leader can lead AI transformation. So that's the topic we're going to talk about today. And looking into the agenda, of course, we will start with a brief definition of what generative AI is. Then we will look about leadership in the era of AI. And then we will touch about generative AI and strategies. And at last, we will go into how change management also can be a really good support when working with AI implementations. So we're talking about generative AI. So we're talking about generative AI. In the invitation, we said AI, but this calls for a really brief definition. Because generative AI refers to the type of artificial intelligence that is capable of creating new data or content such as images. Those of you working in marketing might already have been experienced within that area. As you can see, the pictures from today will also be generative AI. generated by AI. It is a technology that might produce something that humans before have done. So we are super excited to walk you through this and talk even more about generative AI throughout the presentation today. And why is this such a big hype, Marie? We have been looking to it. And actually, you can say that generative AI is the big new new structure of the structure of the world. And looking back, and those of you who are just as old as me, maybe, would also know that, for example, the compact disc was also an element that really changed our way of both working, our way of sharing data, our way of engaging in the world. We also saw how the internet was really also impacting the way we were interacting, the way we were collaborating, across national. So we see that generative AI is here to come across national. So we see that generative AI is here to come. You actually can put it aside, or people are putting it aside with a general purpose technology. Technologies can affect the entire economy. So it is really a thing. And that we also see with the organizations that we engage with. So maybe to kind of underpin this statement a little bit around generative AI being the disruptor of the century. We're putting it and comparing it to the emergence of the internet, maybe even bigger than that, as you're saying, general purpose technology. I think it's worthwhile to just take a little brief look at the last 10 months, because a lot has happened. It's been a very, very exciting journey. And then also maybe a little peek at what is to come. So the next is a bit of a timeline, just to get everyone up to speed. And maybe you get that sense of excitement around how much has actually happened in a very, very short time span. So it all started back in November, November 30th, 2022, OpenAI released ChatGPT, which is the tool that has become the gateway for many, many people to interact with generative AI and really, really put it out there in the public. And it's less than a year ago that this was made available. And not long thereafter, we had Microsoft invest heavily in OpenAI. And that has helped kind of shape the development and it has helped show us or give us some of the tools that are coming today that will be incorporated into the tools that we're using for everyday work. So has been a major implication that they went in and invested this heavily. And then from there on, we had a really exciting spring. So the top here in the timeline is all the generative AI model releases, just to pinpoint a few of them, to say that Microsoft and OpenAI are not the only players out there. We had Mita coming out with some models. Another fun one in February was Microsoft coming with a multimodal model. So when Marie explained that generative AI, AI is generating content, either text images, code, audio, the multimodal models are the ones that can do more than just one modality. So not just text and not just images, but it is, for example, something like ChatGPT with an image component. And the difficult part about doing timelines like this for a presentation in the generative AI field is that things are moving so rapidly. So the whole multimodality thing is actually, rolling out as we're rolling out as we speak in ChatGPT. But it was back in February, we began to see this for the first time. We had the newest model coming out from OpenAI, the GPT-4. This is the one that right now is still the benchmark model. Google announced their release of models. So many, many things happened in February, March, around the whole emergence of these technologies. And at the same time, research and reports began to be coming out to indicate some of the potential that we were seeing with these technologies. So one of the earliest ones, it was a study from MIT, where they were looking at the productivity effects of using generative artificial intelligence. And they found some really amazing numbers, a 40% productivity increase, a 15% quality uplift. So all around this narrative in the beginning or in the spring of the excitement and the possibilities of these models, that also kind of let bring up to a lot of fear, a lot of uncertainty. I don't know how many of you remember, but back in March as well, we had around 1000 signatories joined together to create or publish this open letter. It was bannered by Elon Musk to cause for a pause in what they call giant AI experiments. They wanted a pause of six months, so that we had a chance to catch up to the implications of what this technology will mean for society, for our organizations, and this didn't happen. This didn't happen. I think it was two weeks ago that marks the six month kind of mark from the from the publishment of that letter. And we didn't stop the development. There has been happening quite a lot of things in those six months. But I don't know if we're any smarter today around the implications, maybe a little bit, but it still is a very uncertain field. Also in kind of what should we do? It was quite a big story that Italy went on and banned chat GPT. That was also in March. So crazy, crazy spring when it comes to the whole generative AI development. Italy did end up lifting the ban again during April. A funny story around the whole banning of chat GPT in Italy is that I think around 24 hours after they banned the use of chat GPT, a tool was made public called Pizza GPT, made by Italians to enable Italians to access this new technology. Just a fun indicator that once you put this technology out there, people have a chance to actually work with it. It's very, very difficult to roll it back and then put limitations on it. Crazy spring not there stopping the development. We also saw some of the bigger image models are being released and upgraded. And in June is also when we saw McKinsey come with a very big report around the economic potential of generative AI. This is the past. This was, as I'm saying, very much in the spring. And I think a lot of people have hypothesized around whether or not we've kind of reached the peak of the Gartner hype cycle. I know I've been talking with a lot of colleagues about it, but we're still seeing a major development and big releases coming out. We're seeing investments from big players such as Amazon. They're kind of doing the same thing as Microsoft did back in January, going out and investing heavily in a competitor to OpenAI. We're getting new and better models being released all the time. And finally, we have Microsoft going out and giving us a date for when Microsoft Copilot will be available. So this crazy ride has not ended yet. And I think the timeline here kind of creates the perfect backdrop for both why it is very valuable to start looking into this technology, but also why it might cause a bit of stir in organizations and also in society. Yeah, that was the brief history lesson for the past 10 months. I mentioned very briefly the MIT study around the potential of generative AI and just to deep dive a little bit more on that, because MIT is not a one of the only ones who have been conducting these studies. And during the last 10 months, we've seen a few different coming out. And we were trying to look into what are they saying the different studies around both productivity, but also the quality of the work that is being done once you use generative AI. And this is the median of those numbers. And across studies, we are seeing a 32% productivity increase and a 18% quality increase. Now, of course, these are for certain tasks. This is often for the written tasks. So using text models such as chat GPT, for example. But the fact that there is a possibility to gain some of these productivity and quality numbers is quite amazing. That was the brief, maybe not so brief definition and history of generative AI and the development that we've seen up until now. But that also leads us to how it is being done. And thank you, Marie, for contextualize. And I think it's really important that even though I will now zoom in on the organizational challenges that we have experience with the leaders are in front of, it's also important to state that all those things happening worldwide is, of course, also impacting the way and how that the leaders need to navigate in this really, really unknown territory today. So I would like to deep dive into what are the ideas we see that leaders are facing. Because first of all, generative AI is a new technology and many solutions and many of the solutions are totally new for leaders. They need to conceptualize for the first time. And that is a challenge in itself. Then we also see sprouting grassroots movements. We actually see from our experience that employees, even though you have structured a approach, working with AI, other sub-departments start building their own initiatives while others are conceptualizing in a more structured way. So we really see that even though people work with it structured and intensivized, then employees, they go even faster. And that's really a thing that we see right now going on out in the organizations. Then technology, because it is a hype and many wants to go go on and be part of that hype. We see it really being technology driven rather than business driven, or at least there is a challenge with people only focusing on how can we get that AI to and not even looking at the business opportunities, but also the business problems that we actually would like to solve. Then fear versus excitement from the people that we have engaged within the organization, of course, we can see that really, really big excitement. But under the surface talking to the, with them even longer than we can also see there is a fear related to that. People are thinking about, will I lose my job? Will it also be that I have to redefine my own role? So even though we see the excitement, we can also see fear that needs to be handled in the organization. Then we can see that it's something that everybody knows a little about and somebody knows a lot. Those who have already been working with it, but there are many narratives going on in the organization, different perception about what it is and what it can do for the people in the organization. And then at last, of course, what AI cannot do is to take ethical decisions and employees are reaching out to leaders to get some guidelines on how can I use it and what can I not use it for. So there are ethical considerations attached to working with AI. So all of these perspectives, if we look at it from, you can say, from a more strategic point of view and a relational point of view, you could see those on the left side being linked more to the strategic part and those on the right side being linked to the relational part. And Marie is going to contextualize it into the strategic part. And later on, from a change management perspective, we're going to link it into how the relation elements could be improved and also navigated through change management and leadership focus. Yeah. So as you're saying, let's take a deeper look into some of these pain points that we have identified when we're out there and looking at these AI transformations that kind of calls for for the strategic angles. So as you're saying, it's very, very difficult for many leaders to set this direction in this very unknown field and looking back at the timeline, this very rapidly evolving field. So the generative AI strategy kind of helps us set that direction and alleviate some of the pain points that we're seeing on the left. So what we've seen out there are two kind of different approaches to getting started with a generative AI strategy. So starting with the first one, bottom up. So from the bottom here, what we're seeing many organizations doing is to enable maybe the organization to use tech, maybe they put some tools out there that makes a generative AI available to the employees. Maybe there's a nudge of experimentation in it, then we're crowdsourcing ideas and use cases. Then from there, then they're down selected for further development. Going into the rapid prototyping and then sort of the rapid prototyping and then sort of from that, having an emerging strategy and an emerging operating model around the use of generative AI. Now the big problem with the bottom up approach is the speed in it because everything moves so quickly that if everything is kind of happening in subdivisions and crowded as an innovative movement, a grass root movement everywhere, the speed might actually not cater for the development that we're seeing in generative AI. AI right now. Another problem that we've seen is that when you've used the bottom up approach coming in and implementing, for example, an operating model afterwards can become very difficult if you've already allowed a lot of your employees to use these tools in the ways that they see fit. And then you come with the government and guidelines afterwards, it becomes a very, very difficult change task to kind of invoke some of some of those liberties that have been going on. Now the bottom up here is the most natural actually I would say because as Marie said the innovation is very very very big and and we're seeing a lot of a lot of banner leaders out there knowing that they can use this for something quite valuable so a lot of different parts of the organization will will start to look into this kind of on the other side of this there's the top-down approach which is following a little bit more of our classic strategy approach where it's a clear objective from the beginning then you identify and prioritize the use cases you design the operating model you implement it you test the hypothesis and then you build and scale from there and a bit of the problem with the top-down approach when we're looking into generative AI is that it can be super difficult to set that direction and make a clear cut these are our next three years with a technology that is so new if you don't test it out and if you don't try something along the way so to kind of compromise between the two we're we're suggesting kind of a more transformational approach so still starting with an aspiration what is it that we want to achieve with this technology and then from there identifying the use cases because it should be very use case driven as we're also saying in some of the pain points that we're seeing if it's too focused on technology we're not solving the most valuable business problems or not generating the most value for your company but then what we're going to straight from here is actually experiments so trying it out making these rapid prototypes and getting learnings from hands-on experience because it is difficult as I was saying if we're just saying from the top down kind of approach where we've identified the use cases to then plan out how we should continue from here so instead let these experiments guide you and from the learnings of those experiments get inputs to how to design an operating model how to define the roadmap and define the business case around using the technology. Then from there, once you've gotten even smarter and have the implications for what this technology means in your organization, then you can move on and start implementing the operating model and then test the more strategic hypothesis around it. So get learnings early on from this experimentation. This whole approach is an organizational learning journey because we're putting tools in the hands of our employees and they're supposed to be experimenting and learning from those experiments. It's not just experimenting without a goal. So that's definitely looking into a more organizational learning journey around it. So this was kind of our take on tackling some of these more strategic pain points that we've seen out there in the different organizations. Yeah. So if we jump more to the relational part, then of course, change management can be a way also to tackle some of the relation challenges that we have seen, the fear versus excitement, the multitude of narratives and also the ethical considerations. And just briefly to touch upon it, I would like to say what is at stake knowing from our studies in implement and also our way of working, what is at stake when working with normal change management? It is that there needs to be a clear reason. We know that there needs to be a structured approach. Management needs to take ownership. Trust must be built. Users must be involved early on. There needs to be a clear communication and effective training. And that's all you can say at a general change management level. But how does it really look if we look at it in the context of AI? A clear purpose, meaning and a clear purpose, that is what drives change and motivation. So there needs to be this big why and this how. Not only that we need this technology because it's super cool and it can optimize the way we work, but also we really need to improve this area. It could do something better for our organization or maybe even more for our department. So a clear purpose that is related to also meaningfulness for the people actually working with it. Then there needs to be a structured approach. And by meaning a structured approach, I don't mean, and I've seen examples of that, but I don't mean that we need to do a big end to end, a crazy change approach, but we really need to have a structured approach in the beginning, assessing the impact. Who will actually be impacted? Who will actually be impacted? Will it require new roles? Because having that assessed in the beginning, we can actually focus our change efforts to what is really needed because we see that working with AI use cases are big variety of projects that might not even impact and some does. So having a structured approach for identifying that is really needed when working with AI. Then management, of course, need to take ownership. ownership. And by ownership, I also mean that taking decisions on how at a department or an organization level you would like to use AI. And of course, as we see right now, that that requires also some upscaling because leaders does not necessarily have the knowledge about AI to take that ownership. So that is also what we see right now engaging with the organization. Then trust. And that's something that I really have discussed with Marie because we have had some, you can say, had some interesting findings about that because not only must trust be something that the stakeholders or the employees trust in the overall concept, but also trust in the tool. The tool is a new support. It's non-human support. It's technical support. So trust must be built towards the tool that it's actually a support in my everyday work and that can take time. Then involved users early on, we have seen that the use process as people help to contain, Knowledge data code which will not have to do. So letting us understand the product within Darion and Kad., but how many people move up is appropriate. That's an AL avarling system! of course theirauto.�지 public service can expect to be sustainable with data data data, and the started dashboard in the Powder Management group and developzech blah. Then of course clear communication we can see that many narratives are going on and many perceptions about what AI is. We see a need for building a strong core story that also will be the glue that ties all those initiatives together and makes it easier also for people to meet and engage around a common topic but also having that clear direction. Then effective training as the last thing here is that of course training needs to be effective just like any other IT implementation but what we see is also a need for capability mapping because using a new technology that is also impacting our way of working also requires both some upskilling but also maybe some reskilling coming real that I have to redefine my own job. So we actually see the whole training right now being a big thing not like you come in have a session and then you're out but more that we see organization actually thinking even more strategically how they want to upskill and build those digital learning journeys for their employees. So if we go to the next one looking into what is it that is at stake for a leader because this was a little bit about the change management and how to drive that but there's also things that are at stake for a leader to actually build that AI mindset and work with the culture. A leader needs to embrace the uncertainty. We are working with many unknowns and there will be many learnings as we go along. Sharpen the focus as well. Sharpen the focus as well because of course when you open up that AI box you can find endless opportunities for optimizing and doing it even smarter but sharpen the focus to what is it that we really need to look at from a business and people perspective rather than to the technology and help also your employees sharpen that focus. Accept trial and error is also a thing because as Marie said there will be more experimenting as we go along. There may also be more experimenting as we go along. There may also be failures. So failure is a finding. It's a quite a really really key point here. So failure is a finding. Failure is a learning and that should be part of the culture. Lead from the back. You have so many employees going on and that are really really innovative. Lead from the back because of course they can just as we see they are super exciting and even faster than the initiatives that you're starting with. What you can do is a really big part in your organization but still try to guide them so they are linked back to the overall vision of what you want to create as a leader. Build the energy. As I said they are super innovative and fast but of course there will also be failures so it's really important as a leader to check in with the employees and also to embrace that energy and don't kill the innovation. And then make choices. People is the AI tool. So I think that's a really good. Is it mandatory for my department. Is it just a tool that you could use if you would like to take those choices and be quite clear on how you want to use it in your department because otherwise it will just be totally people working in different directions. So set that and make the choices around how you as a leader would like to use the technology in your area. Cool. I mean this was the overall introduction to how to kind of approach the whole change management perspective in the AI transformation. But we also wanted to give you a chance to make it maybe a little bit more concrete. So to do that we want to look into how change is relevant in the entire use case lifecycle. Because as Marie also were mentioning what we're seeing out there is that a lot of these tools tools are built or a lot of AI tools are built or a lot of AI tools are being enabled and being put out to the organization. And it's only after the development that the whole change management aspect of it is thought into the process. But we have some suggestions for where along in this use case lifecycle it will make sense to kind of incorporate and think a little bit about the initial change initiatives. So we've talked a lot about it being use case driven. Find the the use cases driven by a specific business value and then from there start building and experimenting. And how that lifecycle might look is of course starting at the point where a use case is identified. This is often identified from a long list of use cases. It can be crowdsourced from the organization. It kind of using all of that innovation that is available out there. And then from there it's about prioritizing the use cases. And this is the first step where we're seeing a change perspective coming into play. Because this is where we're kickstarting the communication. We're kickstarting it around what is it that we want to achieve with this. And what is our goal with it. It's also as Marie was mentioning before being aware of the uncertainty because it can be that these things are moving so rapidly that the communication in the beginning will have to change. But it is a very very good start to involve employees already from this point. From a use case perspective, after the prioritization, we would go into a maturation of the use case. This would often include finding the funding and designing it such that we can start building. At this point in the maturation is where there could be a benefit from doing both impact and benefit mapping of this particular use case. But it's also the first point where we're building the management ownership around the solution. Who is it that this use case is going to be affecting? Who should be owning it afterwards? The maturation is also the things like what data is available? What is the technologies? What processes will this be part of? Kind of scoping that maturation of the use case. And then from there we can start building our proof of concept. And in this technical proof of concept, we're testing whether or not, for example, the large language models can solve the hypothesis that we're using. But from the change perspective, we can go out and start making the change readiness assessment. How ready are the end users to receive this tool and to use it once it has been developed and ready? If we do prove the concept and we're seeing that the use case is a valuable use case, we would go on to building the minimal viable product. And because this is the first time that the users are interacting with the tool, this is also the first time that we're starting to really build trust around it. This trust is also put in to the communication around it being very open that it is a minimal viable product and that it can be kind of aligned afterwards to make it even better for the users. Before going to production, we would probably have some early user testing. This is, of course, one of the elements that Marie were talking about with involving the users early on. And then we would go and deploy the use case in production so that we have the tools that we're using. So that we're using the tool available and out there. Once it's deployed, we would have some training and onboarding of the employees. And this is a part where we have to know what is the reason for doing this training? What is the reason for implementing this use case? And it is where our clear reason for change comes into play. Among other places along this line, but also here specifically. And then, of course, this is the training of the users. It's our whole upskilling or reskilling in the light of what this use case will be able to enable. And then the use case kind of goes and lifts out there in organizations. And here we go into a kind of continuous monitoring of the use so that we can make sure that it continuously has the quality that we want it to have. This is where we have some governance around it. Who owns it? What are the different rules? Before going in here, you can do the change readiness assessment again. So kind of look at the looking at, okay, now we've trained our users starting to onboard them. Have we moved this kind of lever? Have we made a temperature check that has changed since our first change readiness assessment? And then while the use case is out there, it is very, very much about capturing the learnings that you get. Those learnings are also the ones that will help you build the trust into the tool itself. And then with some of these AI use cases, the last step in the lifecycle is actually the possibility of decommissioning the use case. As I was mentioning with the rapid evolvement of this technology, there will be new solutions coming at a very, very rapid pace. So to a higher degree than what we might be used to, there's a possibility that we should decommission the use case, take it out of play because there's something else that has come in and is more valuable for us. That would usually involve removing the solution, removing the data just from the technical point of view. And this should be kind of in mind when designing that it is okay, that this might be the end goal because you still have all the learnings along the way. And the tool that might come in and replace it can still benefit from all of these change initiatives that has been going on. So that was to make it a little bit more concrete in building in a specific use case in the organization. But that was our way of trying to highlight some of the pain points. Yeah. And zooming in on the pain points. But Maria, I would like to just wrap it all up and just be sharp on, okay, what are the key takeaways that we would like to highlight from this webinar? So summing it up all up, there needs to be defined a clear aspiration. Always have a why and a what so people know why you are doing it and how and what you are doing. Because it should not be something that only are for people or really heavy data scientists working in a small corner. It should be engaged or also the organization should be involved knowing what is at stake here. Then be use case driven and ensure that tech are solving business problems and not that tech is the element in itself. And then of course, let the experiments guide your direction. Let failure and learnings be your new friend in the organization. Because we are at the unknown territory. And so the key reason to work and navigate in this area is to experiment and do shorter experiments, integrate the learnings, share the learnings, and let those people who are working with it themselves, not in a structured initiative, and those working on the strategic initiatives also let them meet and share those learnings. And then of course focus on continuous communication, engage with people and also create that strong call story that will be the guiding star and the glue tying all those initiatives together. And then change management must be part of it from the beginning so you already at the start point know or at least have some indications on how it will impact the organization from start on. And then of course the experiment and then of course the experiment and then of course the experiment approach sets some requirements for the organization and the mindset. So it's really important also as a leader to work on that mindset and the experimental culture that it is living in order to succeed with all these initiatives working with AI. So that was our way of trying to wrap up kind of this information overload that we gave you. We do know that there's a lot. And Maria and I have had many very, very fun discussions around what we're seeing out there and trying to kind of incorporate all the learnings that you have from the change initiatives and the change world also leading with AI and some of what I'm seeing out there in the whole generative AI field. So hopefully this gave you some food for thought and it looks like we do have some time for questions. So right now we would like to open up the chat. I think some of the questions have come along the way. So we'll just get a chance to look at them. But if you have something that you want to ask us, we have around five minutes left so that we can try and look into it. And maybe I can say because I've seen a lot of your writing, will we get this presentation? And yes, you will. Yes. Good. There's a first question here from Matthias who's asking, how is the public reacting on AI? Is there trust? And can the use of AI in a company be used in branding? So what I think you're asking here is whether or not it's actually a competitive leverage to have AI as part of your organization and the way that you're working with these technologies. So just to kind of answer maybe the last part of this question is that yes, definitely. We are seeing that especially the younger generation is, as we know from other technologies, adopting to this quite well. And they will be expecting that organizations have it available when they finish, for example, their studies, because it's already there at the universities being used actively. And then in terms of the public's reacting to AI, I think it's a difficult question because if you just look at the news, there's often all the kind of horror stories, but that's also what the news are there for. Sometimes it's focusing a little bit on when it goes wrong and that can help shape the narratives. So if I look at the newspapers, I would think that the public would be quite skeptic around it. Everyone I meet, it's a different story. Almost everyone is very, very excited about these potentials. Okay. What do we have? What are the key things to consider when training teams in AI? One of the key things to consider is especially to mapping out the capabilities. So what is needed? And that is not only focusing on the tool itself, but are there also some basic IT knowledge, AI awareness, knowledge about coding that they need to have? So my recommendation would be not only to focus on the tool in itself, but also use and engage with users to understand and really get to know what they need to have. know how you should design the training for them. And then also let the AI tools and the digital way of working be part of the training. Of course, can they work with gamification? Can they do small prompts that they send to each other? So make it engaging as well. Then there's another one here that is, what is your approach to easily raise awareness of the use of AI and its business case to identify use cases and generate ideas with end users? Some of the very successful interactions that I've had in this field is actually going out and doing big brainstorms in the organization through workshops. So in this workshop, it also allows you for the possibility to give a brief introduction to what is generative AI. That kind of gives us all the foundation of how this technology works, but also what are the main risks and challenges associated with using something as, for example, chat GPT or some of the image models. And then from there, taking a structured brainstorming approach to gather and capture all the ideas. Because one of the most exciting things about this technology is because it is so generally available for everyone, we're seeing a vast amounts of different use case ideas. This is not a technology that kind of springs out of the data science department that has kind of their world to relate it to, but we have everyone interacting with it. So a broad kind of involvement from the involvement from the organization to gather and capture these use case ideas is something that works really well. Then we have Veronica asking, what are your thoughts around starting small with narrow use cases to understand more about this technology versus starting big with transformative use cases for the industry, given the maturity of generative AI? Super good question. Yes. And then you can add on because of course, as we said, there's also a need for building trust in this, there's also a need for building trust in this. And you need to learn your organization, how will we work with that? So when you start small, you also sharpen your focus on how it could be used. And then you get some learnings on how it really fit with your overall value proposition and way of working in your organization. So it's not that you should start small over a lot of years, because of course, you have to jump on the AI train, but start small with a smaller scope and then get the learning and then you can scale it bigger also to engage employees and to ensure that trust is built toward that. So it's not too big, but testing it out at a pilot or a smaller scale would be recommended. Yeah. And I think maybe just a comment on that as well is try and investigate whether or not there's a possibility to do both. So you have the opportunity to do the narrow use cases and get some of those learnings for it. But you should, of course, keep in mind that they can be quite transformative. So it would also be a a little bit blind if you just go in and only incrementally look at the small use cases along the way. So have it have it as part of kind of the the way that you think this is this is going. And then we have one here from Lawrence asking how iterative should the change management process be approached with AI development at rapid rate during implementation of AI into an organization? Now I'm looking at you, Marie. Now you're looking at me. Yeah, of course, it's going on rapid. And that's also the difficulties because of course, you need to communicate and prepare and that might be a slower pace. But but as we see, it could also be having identifying some of the early adopters, identifying some of the really innovative, innovative people in the organization, be those being the new digital change ambassadors that could also ongoing drop all those things going on under technology. So that's the rapid side and then still have the communication that is at a normal pace. So try to build in that. But of course, it's also a key point not to to make it too big for the employees that they are disturbed all the time. So it can also what we also are looking at when working with change management is also to make for us that invites people in that are really interested in the change that would really like to know about everything, but also give a pause to those employees that are not really that interested in the same. And that's totally fine. They need to know when the tool is launched. And then we have those who really want to be in the front. So also what I would recommend is to do proper segmentation. You were doing that with like the last seconds counting down because we are running out of time. And this was our 45 minutes. Yes. It was a lot of fun sharing some of our insights with with all of you. And if you have any questions, do feel free to reach out. You see our email addresses here on the screen. We will share the slides with you afterwards. But thanks for tuning in. Thank you. Thank you for tuning in this morning.