Good morning and Good afternoon because we see that several of you are dialing in from time zones later than I am Gitanjali Ponnappa, and I have had the privilege of having a 25 -year career in being a manager of change. And most recently, I am nerding out on what AI means for leadership. So, my name is Pascal. I'm a huge nerd when it comes to artificial intelligence, and I have a deep passion for preparing people for change, especially when it's digital change. I'm looking forward to this session with you. Before we jump into content, we'd like to tell you a little bit about us. We are Implement Consulting. We want to leave every organization fit for humans and fit for the future. We have developed this vision for ourselves, working with thought leaders across the world. We have Nordic origins. We are, let's say, a Scandi transformation consultancy. We have offices here in Switzerland, as well as in North America, in Raleigh, North Carolina, across Germany, and, of course, in Scandinavia. We also want to be your transformation partner. When clients ask me, what is it that you do, I don't say we do everything. What I say is we do six things. This is our sweet spot. And today, Pascal and I are going to cover two of them, the one that says digital and AI and leadership and change. We're going to mush them together and have a conversation about that. Thank you so much. 95% of all AI projects fail. That's what a recent MIT study says. 95%. That's a huge number. Maybe it's even a little bit too big. But it shows a very real issue. AI adoption, AI integration is really hard. There are a lot of challenges involved. Let's break it down. So, first of all, of course, there's a strong human element. There's a lot of fear and hesitation when it comes to artificial intelligence. Leaders need to be aware of that. Another challenge is organizational friction. A lot of AI projects and AI integration in general works across different departments and knowledge domains. So, it's especially important that the different structures and incentives are aligned so Genitive AI can actually scale effectively within your company. And lastly, of course, there's a technical component as well. A lot of these AI projects don't make it out of the prototype stage because they break in the real world context with messy data and differing users. These three elements are all really important. But today, we're going to look at the first one, the human element. What is quite unique about the AI adoption story is that there's a huge pull from the end user. There are a lot of people who really want to use artificial intelligence in their work. Three quarters of global knowledge workers are already using Genitive AI regularly. That sets it apart from other digitalization stories. But, of course, that doesn't mean that there aren't still a lot of tough challenges involved. We are going to look at the three of them in a minute. The first one, the 10% versus 10x. This is a mindset topic. It's about dreaming big. Shadow AI, it's about people bringing their own AI tools to work. That is great, potentially, for productivity, but it creates a huge risk when it comes to data security. And lastly, of course, there are still people who are not using AI at all or not using it for their fullest potential. We will also address this challenge. Let's dive in to the mindset issue. 10% is the enemy of 10x. This statement comes from Jeremy Adley, a Stanford professor, that we had the pleasure to talk to a couple of weeks ago. He argues that we spend too much time thinking, how can we be 10% more efficient? How can we save 5% of resources? These are important topics and these are important savings, but they don't fundamentally change the game. Jeremy argues that we should dare to move beyond efficiency and ask ourselves the big questions. What can we do now with AI that was simply impossible to do before? Or really take some time to reflect, what would a 10x leap in our work look like? How would it change? Look at shadow AI. So what is shadow AI? So I have my own personal chat GPT subscription and I use it at work. That is shadow AI. Whereas in implement, we actually have what we call IMGPT. I should be using IMGPT for everything, but I've now brought my own tool to work. In the old days, it was the bring your device to work. You brought hardware to work. You brought your spare phone. You brought maybe a Mac because you didn't like the laptop that was being used. But now what you're doing is you're bringing your software to work. You're bringing your AI to work. And as Pascal said earlier, it does create a risk when you are sharing sensitive business information. Let's look at the next one as well now. The non-users. So we have seen there are a lot of people who are using artificial intelligence a lot. But there's still many who don't. And even of those who are using it on a regular basis, they don't use it to their fullest potential. They use it just for the basics like writing emails, summarizing texts, and using it as a Google replacement. Don't get me wrong. Those are all important use cases. I do all of these things daily. But it must be only the start and it can't be the end. This is an incredibly important topic for leaders because this creates a tension not only in teams but entire organizations. You have people who are rushing ahead, who always know the latest tips and tricks with AI and their productivity source. But you have others who do not touch the technology and they stay at traditional productivity levels. It is up to leaders to bridge this growing skill and also productivity gap and give everybody a chance to learn how to leverage this new technology effectively and get on board the AI journey. Everybody has a framework. Well, we have a framework too. Ours is around how do you improve adoption of AI. And our framework addresses these three challenges. We just talked about non-users. We talked about shadow AI. And we talked about how you need to have 10x thinking and not just 10% efficiency. Now, in our framework, let's look at the very first one around leadership. So maybe even a year ago, as a leader, you could say, hey, IT needs to handle this. IS needs to handle it. But now, with AI easily available, even with not company-sanctioned products like maybe my personal version of ChatGBT, it has now become an issue around power and permission. Let me explain that a little bit. What I mean by that is it's not about should I use AI. Now it's become about how do you use AI. It's no longer a question of should. It's a question of how. And the power comes in with you as a leader giving permission to use AI. Let's talk about what these permission boundaries, this box, might look like. A permission boundary, the way we see it, has three elements. Element number one, what do you use AI for? So you explicitly state what AI can be used for. As an example, you as a leader should encourage your team members to turn on recording and transcription so that you get meeting notes. And we still have colleagues who are worried or clients who are worried that our data is being stored. So finding that balance of when you can use AI. The second one, when do I use AI but with guardrails? So defining what those guardrails are. The guardrails, let's say if you're in banking, the guardrails are, please don't use AI if you need to upload customer data. That would be a guardrail. And then the last one is, what do you not use AI for? Right. So you can list out those kinds of boundaries. Another example is about what you do with that boundary. How do you reposition your authority? And you do that by encouraging the use. You do that by talking about how a product developed with AI will be reviewed by you. So what will you refine? What will you reject? That is how you reposition your own authority when it comes to leading your teams in the adoption of AI. Let's look at the next one. This one is around training. And perhaps we should have relabeled that and called it skill building. Maybe we should have called it habit making. Because training is essentially perhaps the wrong word. Because using AI now, the proficiency is more like a skill and a habit than it is about just being trained. So training needs to move from this event-based training. Oh, I've attended a webinar. Oh, I took a micro learning session on something. It needs to move from that into what we like to call workflow-based learning. So let's take an example of that. Anchoring AI to just maybe two core workflows just to get you started. You don't need to bite off too much. As a leader, you can make recommendations to your team by saying pick two core workflows. Now, what does that mean? So I take an example again. A simple workflow could simply be usually it is the junior -most person on the team who has to take meeting notes. Well, get rid of that. Have AI develop the meeting notes. Now, think about how that fits into the flow of what the work is being done. Another example could be, let's say, month-end closing. Pick one microprocess within month-end closing that is, let's say, about gathering data. Can that be done smarter and faster by doing that? Let's take another example now. Create these coaching circles. A coaching circle is, let's say, where Pascal and I, we meet once a week. There are no slides. There's no PowerPoint at all. What we do is we discuss, hey, in our work, it's required that we do three LinkedIn posts a month. How can we make this painful, annoying thing that we have to do faster but also smarter? Can we come up with content that's attractive, content that people want to read? Now, that is a practice loop. We sit down every few weeks together. No PowerPoint. It's real world. Now, I'll let Pascal talk about the next one around communications. Thank you, G. So, communication is a huge and important topic when it comes to successful AI adoption. We have seen in our work with different organizations that communication around AI is often either quite technical when it comes from an IT perspective or, on the other hand, it is really visionary. Dreaming big. Dream big. Now, we've learned from Jeremy Utley that dreaming big, that's important and we should do it. But it can't be all that we're talking about. Because if your communication is too focused on the hype and amplifying it, then this just creates more anxiety and uncertainty in your organization. People will wonder, okay, if this AI is so great, what do you need me for? Will I lose my job? So, instead, you should aim your communication at reducing fear and providing certainty. So, how can you tackle this as a leader? There are several actions you can take for that. First, replace these visionary statements with priority. What does that mean? It just means being as specific as possible and talking about what will change for the different roles involved. It can be very simple. For instance, for a project manager, it could mean that in the future, they will need to spend a lot less time finding meeting slots for everyone because AI is going to do that for you. It's an extremely basic example, but it makes it tangible. It shows the person, okay, what will be done with AI in the future. But when communicating, don't forget about the human. Don't just talk about what will be done by AI, but also what will stay human. This helps people to see themselves in the future that you envision for them. Another action you can take is to make experimentation visible and normalizing it as well. Pick a story, either from yourself or from a colleague who has consented, and show how you use AI. And don't just show the finished perfect output, but show the whole process. Show the iteration, show the prototyping, show how you address hallucination and how you really get to an IS result. This is not just about creating a learning opportunity for others, but also about creating a safe space where people feel like, all right, not being perfect AI, that's actually normal. And sometimes you have to really dig in and actually work with it to get something nice. This last point really ties in to our next element. Culture. Culture is also hugely important when it comes to AI adoption. In companies we've worked with, who actually integrated AI really effectively, we have found that very often the culture is actually not even that AI positive. And sometimes there are strong AI critical elements. But what all of these organizations had in common is that they're error tolerant. It's okay to make mistakes. It's okay to fail. And it's okay to talk about that. Shifting a culture or norms is really tough. But as a leader, you're in a great position to notch your organization in the right direction. What can you do? Well, you could institutionalize rough draft first norms. What does that mean? This is again about being explicit. It's about writing stuff down. For instance, in a working agreement or in a team charter, state explicitly that it's okay to share first drafts that you created with the help of AI. You should move the norm from share only perfect output to share early and refine together. This leverages the full prototyping capabilities that are provided by AI. The first action you can take as a leader is to publicly deconstruct imperfect outputs. You can do this in team or in leadership meetings. Every once in a while, look at an instance where you or again a colleague tried something with AI and it just didn't work. Maybe there was a lot of hallucination. Maybe you actually trusted the output, but it turned out to be wrong. Again, don't just look at the output. Look at the process. What did you do? What were the prompts? What were you trying? What didn't work? Open up the space for everyone to ask questions and to contribute with their own ideas. This will not only create a learning opportunity where people who are maybe not up to speed when it comes to AI to learn something and be inspired, but it's also again about normalizing that it's okay to fail when working with AI. And it's not about failure, but it is about learning. So please don't forget about culture. It is an important aspect. And as a leader, you're in a great position to shape it. I really like that idea of fail and it's okay to fail because when we fail fast, when we realize that there's a hallucination or we realize that, oh, this doesn't actually look like the way that I wanted it to look. Then we make space for improving the product. And I would like to call that failing forward. So you fail forward with something like that. That actually ties in with the next aspect that we want to talk about with you is how do you manage performance? What does performance management look like in the age of AI? Well, for one, I think we should address that. AI is scaring people. There's fear. There's fear that comes from I used to analyze this report and it used to take me five days. My team was used to me taking five days and now I take two. So my team is probably thinking, what is she doing for the other three days? There's fear that comes from the use of AI as well. And if we acknowledge that fear and we move people from the production, the output, as you called it, to the process. If we can get our teams to lead by process and not by just output. And we also always classically say also outcome. So let's take an example for that. You must reward smart reductions. So a smart reduction is the example of it used to take me five days to produce this, to analyze and produce this report. Now, the analysis and the production has come down to two. I should reward that. So what might that look like? Like I say, in real life, what that might look like is you actually reviewing the product, like Pascal said earlier, together in a team environment and talking about how the product is. Talking about how the process was shortened, smartly, talking about how the process perhaps was also improved because of the use of AI. So you reward that behavior and work towards making people feel less uncomfortable about the speed that they're gaining. And right now, acceleration speed is the 10 percent efficiency. I know. But once we get comfortable with the 10 percent efficiency is when we can start thinking big, when we can have vision, when we can ask AI, hey, tell me, is there a better way to do this? So we change the quality of the quality of the quality of the quality of the quality of the prompts that we use with AI. Let's look at another example. It is critical as a leader that you expect your fellow leaders to also have AI fluency. Now, AI fluency. I give you a negative example. I've heard people actually say, I can't get myself to use AI. That should be unacceptable on your leadership team. You have to encourage the use of AI. You have to get your peer group of leaders, people who sit in a steering committee, people who lead other people. You need to get them to have the language of AI. Now, just to be clear, I don't mean technical speak. Right. So I'm not talking about the technical language of use this and don't use that. And this is the framework behind it. And these are the models. Not that. What we mean when we talk about AI fluency is the aspect of judgment. How do you know that this product that has been developed now with AI is actually going to have impact in your organization? How do you look at that report now that your controller has prepared for accuracy? How do you smartly check it? Human check. Yes. But how do you check it for accuracy to get rid of hallucinations, to get rid of wrong sources, perhaps to get rid of old data or faulty? Let's let's call them faulty conclusions. That is the role as a leader is to develop the fluency around better judgment. It's about helping your teams develop the language required to say things like, yes, I use AI and this is the output. Yes, we can review it together. And yes, these are the three things that we look out for when we review, refine or then finally reject a product. So going forward from a performance management perspective, it's good to not just reward speed, but to reward smart user AI and to ensure that your leadership team has that fluency. You remember we told you about the wheel when we started talking a few minutes ago. We are now almost through the wheel and we are down to the last bit in that wheel, which is about the user's experience. So when I say user's experience, I mean me, I mean, Pascal, I mean you. When you log in and you have to use AI, what is your experience? I want to give you a little example. I had a friend who called me yesterday. She's also a client of ours, a client friend. And she said, you know, I'm really struggling with articulating this message around data fluency because I run a data project. We talked through it. And then when we were wrapping up, I asked her, hey, by the way, have you asked a I to coach you with your thinking? And she went, really? I didn't even think I could do that. And so we talked through how do you get a I to give you the prompts to help you think. Now, those are the moments that you want to design for when a person has not even thought of using it. Now, a adoption does not fail because of people like Pascal and me, and it also doesn't fail because of people like you who dialed in. We already know the value. It fails because of these people who haven't thought to use a I. I. Let's build out that just a little bit more. You need to go search for these hesitation points. Hesitation points, for example, are I am really worried about privacy. And so somebody stops using it because they have an overarching imagination of data privacy, whereas the task that they're doing perhaps has nothing to do with data privacy. I give you an example. You now have a steering committee meeting and you have received a 500 page document that you need to digest. Maybe you don't the document because you're afraid of data privacy issues. But instead, what you do is you ask the A.I. to say pretend you are now on the board or in the steering committee. What might be the three questions that I need to prepare for? And then you read the 500 page document keeping those three questions in your mind. It might improve the quality of your preparation. So these hesitation points, search them out so you can address them. Let's look at another one. Design for the reluctant, those who don't want to use it. I mean, if you're designing for us, we're going to use it anyway. Right. But design for those who don't. In the old world, I don't even know how old that world is, but in the old world, we used to call it user acceptance testing. User acceptance testing would be as an example. You would have a little camera function on the person's laptop and you would see where they are clicking and what they're doing. So they are in procurement and they have to procure 19 hammers for this site. You follow them along and you see where they click. Another example of user acceptance testing was, let's say, in safety training on site. You stand and you actually physically observe somebody what they're doing and then they take out their handheld and they have to punch something in. That's observation. With AI, the same thing happens is you have to see where do people not take out the handheld to log something. It's the AI equivalent of that. Another area would be in user acceptance testing. You see somebody just abandons what they're doing and goes back to, let's say, their Excel spreadsheet instead of using the finance checkmode that you've built. You notice that people go into manually writing the 10 emails and just changing the email ID. You notice that they've abandoned the use of co-pilot to do that. Those are the moments that you want to design for. With that, we've covered the six aspects of how you could address the first three challenges we discussed. The challenge around efficiency and 10X, big thinking. The challenge of having these shadow users. And now the challenge of non-users. Pascal, I believe that you want to tell us about the AI Academy. I do. So now you've heard a lot about AI and leadership. But we offer AI academies where we go into a lot more topics from AI strategy to security to agents. So if you're interested, our next AI Academy is going to happen on the 6th of May here in Zurich. But please note that we also offer tailor-made academies for companies. So it suits your needs perfectly. Perfectly. And with that, to bring it all together, as we said, we have an approach. We've given you a little sprinkling, a little taster of the six where we talked about leadership. Training, perhaps we don't want to call it training, but we want to call it habit making. We've talked about communication. Culture, the ways of working, making it legitimate to use AI. Performance management, where judgment is the new KPI. And we wrapped up by talking about designing for these moments of hesitation to convert them into moments that matter. Thank you for listening. With that, have a very good morning. And we look forward to hearing from you. And thank you, Pascal. Thank you so much. And have a great day.