Is AI building your competitive advantage, or giving it away?

Hello everybody and welcome back to this episode of AI Watch. Peter is back with us here today, and we have an exciting agenda ahead of us. Since we talked last, the world of AI has done anything but stand still. so there is lots to talk about. So great to have you back, Peter, and welcome. Awesome, thank you. Thank you. Thank you. Always a pleasure. You've been busy since we talked last, and one of the fruits of your labor is an article, talking about something that for some could seem a little bit futuristic. Can you just put a few words to what that concept you bring to life is? Yeah, so the concept is really about the knowledge created in organizations, which I've called AI cognitive sovereignty. Not necessarily minted on national sovereignty, but it's the sovereignty of the business and its processes and what makes businesses unique or organizations unique. And the challenge we have going forward is we've been, in the last five years, very focused on managing the knowledge that we as individuals put into a system. And we have attempted to manage this, policy has been put in place by organizations, but It's very focused on the individual. I use Copilot, I use Claude, and I put the information in there. What can I put in and what can I not put in? And that's an ongoing issue that I think we'll see going forward for several years. But with the explosion of agents, the implementation of these agents and workflows, what you now start to have happen is that the large language models capture your organizational knowledge map. So that all the things that happen between individuals working together. That can be defined as culture, it can be defined as innovation, it can be defined as agile business processes and decision environments. So all those things that make a Harvard business review case this organization unique, because it it it does things better than than others that now gets encoded in a system and that is what I would like people to at least think about when they start their journey of deploying agents within workflows and having AI take over some of that work, because: Do you own that? Because Historically as an organization things like culture, like innovation, like agility, have been your characteristic of what makes you unique as a business and also enable you to compete, or if you're in the public sector, enable you to differentiate the service to citizens. And so you want to keep control of that. You want to own that. And and that's what I think is something that we need to highlight more. Okay. So I think for for some that future might sound plausible, but conceptually quite far from where they are right now. Let's say I'm a manager, I just deployed Copilot pilot or enterprise version of Claude or ChatGPT GPT to my employees. So what am I doing there versus what you just described? Well, you're in fact doing what I just described in that second scenario already now. Because all the people working with either Copilot or Claude, they are building agents, agents that change dynamically, agents that do work that is team-related work. This is what we, for a long time, had called shadow AI, is that that that shadow AI now is not just about the individual, but it's about the business process and knowledge that exists in a team, in a department. And so you're building this knowledge map of how you work and what makes you unique, even if you today have just deployed Copilot. So you're doing it, and you're doing it now, and it's fine. I mean, this is how we use the technology, but We have to go in with our eyes open and understand that what makes your organization unique now exists in a system that is owned by somebody else, unless you do something actively to ensure that they don't, or at least that you can port it over to another system so that your unique business doesn't become dependent on a particular software provider. So So Boiling it down, this future that you're painting is happening, whether we like it or not. The question is really whether we as as leaders of companies or departments are going down that path with intent or whether letting it happen coincidentally and hoping that we won't get stuck somewhere in the middle. Yeah. So Knowing is much better than not knowing. As we also know with what what is happening with shadow AI is we shouldn't necessarily ban users building their own agents. That's the whole purpose of this. But we should know about it. We should at least monitor it centrally, also because regulation requires you to do so. But awareness at least leads to the fact that we can then start to separate, what is unique, and ultimately make that knowledge base, the the memory, the the system architecture, the data architecture of that knowledge portable. Because that that should be your mindset and really the main question you have to ask yourself, can I move this? What happens if I want to change provider? And and that is something you need to think about now. Because thinking about it three years down the road, it becomes much more difficult to disentangle. Okay. So if we again take it down to that manager that that just started out with one tool, whether that's Copilot, whether that's ChatGPT GPT or or what have you, the decisions that that person will have to take at this point, if if I understand what you said is okay, know what it's being used for. So there is an element of just having locking or whatever, so I know what intelligence am I building in this system. Yeah. But you also mentioned shadow AI. So I pick one system, but but I need to know what happens in the in the peripheral. Are people using other systems as well? Yeah. Or just free solutions of the system that you're using. I mean, it may be that your corporate solution is ChatGPT GPT, but you know, thirty percent of users actually use the free version. and so this has to start with awareness and it is an executive responsibility. So when when we start to deploy technology around artificial intelligence seriously in organizations, beyond essentially a personal productivity tool, then we need to start to think differently about that technology, and we need to ensure that we are in control. So let's say I do that. Let's say I deployed Copilot. I also scan my perimeters. So I see yes, they are using the free version of something. I can close that down and take some active choices to channel most of the information that I'm sharing with these models into tools that I'm comfortable with. Now, what we're seeing in some organizations is then you begin to move from this, as you describe it, individual productivity. I work with AI to get this task done, I work with AI to get that task done, into more collaborative work. Yeah. Either between people or agents. I suspect that this takes a different form, because now the information is not the document I uploaded, can you talk a little bit about that? Yeah, so so so that is where that shift happens from a focus around personal productivity and what individuals do with the tool to either teams, groups, departments, or the entire organization starts to work. So the work process is designed around the agents that operate. And now I'm encoding what makes me unique, not as an individual, but as an organization. The organization is a living mechanism here. And the organization is unique on a set of different dimensions, some of which I mentioned could be agility, could be innovation, could be your culture. And that I need to manage. If I if I lose that, what do you have left as an organization that differentiates you, from your competitor? And so if all those aspects become freely available in a system and I cannot move it. Because the provider does not allow full portability, not of an individual's activity, but of a team and a group's and a department's activity, then we may have a problem. And I'd rather want to do think about that now than three years down the road. Yeah. Yeah, because it feels like a almost like a reverse because if if you look at some of the the greatest companies throughout history, I I think about someone like Spotify that really rethinks the approach to product development and works in agile teams and so on, that whole idea was very well anchored in the company and became a competitive advantage in itself. But what you're describing is now I'm moving a lot of that dynamic dimensions and way of thinking into something digital that now lives outside my company in principle. Or at least lives in the system that you are using and cannot be taken and moved from Anthropic over to a Mistral solution. Yeah. So I I want to be independent of these environments when it comes to what makes me unique. Yeah. And you're absolutely right. Spotify is a great example. You can take Apple. I mean, We all know the the way Apple designs things is a uniqueness. There's a whole department that does that in Apple. They're using these tools as well. Are they in fact giving away what makes Apple unique? And can they not port it somewhere else? I don't know, but you know, you can go through every industry, yeah, and even in the public sector, you can go and look at, you know, here's what makes us unique. And so for everyone, we have to think about that uniqueness and we want to make sure we control it. That's what leaders do. and and so This is why there are great companies and there are mediocre companies. True. But if if the blueprint is digitally available then all of a sudden the distance from mediocre to great could be shortened. Yes. But then I guess others have to be even smarter. You could play devil's advocate on this and basically say, well, actually it's okay. What will then happen is just competition will then accelerate because the moats that have existed for certain business models disappear, simply because they are now part of a fabric that runs an organization. So there is a scenario that would argue counter to what I've said, that essentially this becomes encoded in all of the systems we use, and therefore the way businesses differentiate themselves has to change. Now what that change is, I'm not sure I know yet. Interesting. It sounds like a future of change, no no no matter what we're we're talking about here. Okay. So if we then look at this step by step and and now we're getting close to the to the picture that you actually drew in in the article, what is it that you say to leaders when you are advising them on on where they are now? What are the things they have to do? You talk about this concept of exit. What are the questions you have to ask to entertain that? I You need portability in the uniqueness of your business. So you need it to be portable from one provider over to another provider. This is the knowledge map that differentiates you, the uniqueness that is you, and that needs to be portable. And So the first thing you do is ask yourself that question: Is this portable? The answer to that question, will outline what you need to do. Because I I'm certain that for most organizations right now, there will probably be two answers. One is I haven't got a clue. And the other one is not, it's not, because we haven't thought about it. This is the main question that has to be asked. Is it portable? The question then is who should ask this question? And I think this resides resoundly with the CEO and the board, or you know, the the executive responsible for the organization. They are the ones that need to think about this. They will then assign operational responsibility to certain parts of the business. This becomes a business unit leader responsibility. This is perhaps what makes you unique as a business unit leader is I can actually manage the uniqueness in the organizational knowledge map in a way that makes us more competitive than others in an environment that is governed by AI agents. Okay. So I've had a few conversations with with CIOs in in the past sort of sovereignty era here. And the term cloud exit seems to get everyone a little bit pale. A lot of people have invested decades of their career getting into a cloud and now they're being asked what does an exit look like? is it Is it the same binary exit that you're talking about or is it is it more in parts? Because the reason I ask and it's a loaded question, we find that if we talk a more selective exit, then people have a an easier time accommodating it. Take what's risky out of a cloud, let's say, or take what's very sensitive out of a cloud. I can do that. Yeah. But not the entire thing. Can the same logic be applied here? Yes. Okay. Nothing is ever binary, you know, or black and white. We we always end up in the gray zone in between. And I think this is also the case as in some instances, like, theoretical example of Apple is that their product development division that is some of the uniqueness of Apple, they want to retain that and they want to own that and they don't want anybody else to get access not to the information but how they work. Yeah. Right. But it could be that the way that the accounting part of the business operates, there is no unique differentiation there, so that's fine. So I think that's how we need to use it. Okay. So if I took what you just said down to let's say I'm I'm the CEO of a of a manufacturing company. I would look at my business saying my R and D and perhaps my production is is where my secret sauce is, this is what makes me unique. I'm able to deliver certain things and so on. So in that, whatever I do with with AI and that, if that's agents or giving people assistants this is where I would focus on the portability. In marketing that may be more plain vanilla. I'm not doing something magical compared to the next guy. So here utilize whatever and and perhaps even accept lock-in because the risk is limited. So it would be it would be basically addressing it like that. Okay. Yeah. So I think that that probably puts a fair bit of pressure on having formed a view of that. Correct. Which is why we come back to who should be responsible. And it has to start with the CEO and the board. Yeah. Because you need to basically sit down and say, what is it that makes our organization unique? What are the workflows and the way we work and the way people interact? And then from that, determine what are you doing with that information in terms of using large language models and and basically AI to optimize things. And then operationalize whenever we get a great idea, if it's in production or R and D, you follow this avenue, which is portable and allows us to protect. If it's in marketing, you can follow this other path. And then operationalize that yeah, governance model in in an organization so you have a future where the most holy of your company can be ported. Correct Can be protected and ultimately is sovereign. Which is why the the word portable is the question you ask yourself, is it portable? And your answer will essentially tell you what to do. Yeah. If I don't know, it's because you have no clue what's being done out in the business. So that's your first step, figure out what's happening. If you say, no, it's not portable, but I know where we're using it, then I can now determine which departments are the strategic ones for me. And if you say, yes, we've already thought about this, awesome. Very few have. But the portability question is powerful, right? It distinguishes sovereignty circus from genuine sovereignty. Yeah. And you know, there are a lot of other technical aspects to this, but I think with all things it's really important to make things s simple and the word portability. I think everybody can get that. Thank you for sharing always great insights and I look forward to picking this up again. Sure. Thank you. Thank you. And thank you for listening to this episode of AI Watch. We look forward to coming back in the not so distant future.