Building Your AI Strategy - AUTONOMOUS session

To Phil Philip Deuter, CEO of boi, who will be talking to us today about building your AI strategy. Just as a reminder before we begin, the goal of the chat is to discuss the content of what Phil is going to be discussing. And there is a Q and A function. If you have questions, please upvote each other's questions because I'm going to use that to figure out which questions to pose to Phil. I don't want to see any LinkedIn promotion in the chat. Please just message each other individually. That's the goal, is to really talk about the content. So without that floor is yours, Phil.

All right, thanks a lot, Kevin. So, want to share with you some perspectives on how to set your AI strategy. So I'm Philip, CEO and founder at poi, for those that have just joined, and I want to take you through the foundational perspectives on AI, which for me form the basis of a good AI strategy. Then a framework to actually set your AI strategy, and one concluding thought around leading that transformation in your organization. So these frameworks and perspectives are drawn from the work that I do at BOI across both AI strategy and build, but obviously today mainly focused on this blueprint, which is our AI strategy framework. But more on that in just a minute. To me, every good AI strategy should be grounded in these four foundational perspectives, which I'm going to walk you through one by one. First one is AI adoption will be happening in three waves. The cost of anything digital will near zero, and it's really not just about the engineering of intelligence, but also about the engineering of biology. And then the fourth one is that AI in itself will actually not lead to a competitive advantage in itself. So that first perspective is that AI adoption will happen in three waves, very much similar to how previous foundational and general purpose technologies have evolved and have been adopted. Where the wave one is very much focused on efficiency, doing the current thing faster and cheaper. The second wave is focused on leveraging the new technology to get to better outputs or enhanced quality. And now wave three really gets you into system transformation. And we've seen that with electricity, we've seen that with the Internet, we've seen that with mobile, etc. And now we're seeing similar things with AI. So what I'm advocating for is that in my mind, it's a mistake to just look at AI as a pure productivity tool. Obviously it leads to productivity. There's much more to be gained and it could really change the business model of your industry or change the entire dynamics. Just running through a super Practical example, example in terms of video generation, one of the domains which is in high velocity development in the AI space. Whereas the first wave is really about how can AI help me to do video editing faster and cheaper. And we've seen some companies launch AI generated video campaigns, for example, which would squarely fit into that box. The second wave would be to think about how can it actually get us to better quality. Are there certain types of videos that we couldn't make before that we can now make with AI? And then the third one would be really to think about an entirely new model. For example, could we have real time movie generation for an audience of one, so that each of us would see their own personalized movie, their own movie end, or we would all see our own personalized ad campaigns instead of all the same. So that would be a wave 3 type idea. And the important strategic insight here is that being successful in wave one is not @ all predictor of being successful in wave three. So these are really tracks that you should look at to work on in parallel to one another. The second perspective for AI strategies that the cost of anything digital will near zero, and that that is a very fundamental change. Obviously this follows decades of Moore's Law where the cost of distributing anything digital has already gone close to zero, or at least the marginal cost. But right now, also with Genai, the cost of production of anything digital, whether that's software, creatives, games, music, whatever you can think of, will also start approaching zero, as well as the time to make it. So if we think that through, this might enable almost anything to be generated in no time at close to no marginal cost for each additional unit that we create. This might enable us to make our own Hollywood movies from our own desks, where we select actors like you would select a font to type in a word document. This might enable instant UX being generated in the moment just for ourselves rather than standard UX that is the same for everyone. It might enable real time simulation of of all types of business decisions that are taken either by humans or either autonomously by AI agents. And obviously, in general, if you look at predictions, this might be the year that we start getting more AI agents by the end of the year than actual people on the planet, which will obviously be a game changer in itself. The third foundational principle to really ground AI strategy in is that while all of the talk is about the engineering of intelligence, which is what AI is that I would love to invite you to think about the next decade of really the combination of that engineering of Intelligence with the engineering of biology and first steps in the engineering of biology is what you would see with GLP1 type technology which, which starts to engineer our levels of craving, our probability to get certain diseases and how they evolve, etc. On the other end of the spectrum, obviously you have all of your LLMs which we all know, et cetera, but right now they still live and exist as separate chatbots that we interact with kind of on the side of our lives. And it will get more interesting as they become more immersed into how we interface with the world. For example, my prediction is that in the next three years we will see an iPhone moment or a ChatGPT moment for augmented reality glasses, which have been long overdue and long promised. But I think the next three years that turning point might be there where LLMs might become integrated in our AR glasses or into our AirPods so they become more immersive than the current chatbots that we have. Obviously you have the whole evolution in humanoid robots which currently mainly live in factories, but will in the next couple of years get into our everyday lives and then what? One step further beyond that, which would really start seeing the convergence of the engineering of biology with the engineering of intelligence, is what the companies like Neuralink are working on, which would enable us to integrate AI with actual human intelligence. There are bunch more technologies like live extension tech, synthetic biology, etc. Which I won't get into right now. But this is just an invitation that if you set your strategy for AI to think through the lens of both AI as well as engineered biology, which in my mind will go hand in hand in the next decade, then the fourth perspective to ground your strategy in is that adopting AI, although, and I think that's one of the biggest misconceptions, is that adopting AI will lead to a competitive advantage, where I would argue that it won't because it's very much general purpose technology that almost anyone in the world has access to either for 20 bucks a month or even for free with open source AI models which are very powerful today. So where is competitive advantage going to come from? It's going to be that combination, unique combination that you can put together of the technology and the data, the AI with the operating model, with the people that you can bring together. So it's going to be a unique human AI system that you as an organization can bring to bear, rather than just deploying AI that would get you to the competitive advantage. All right, so let's continue a bit on that and think about some of the core components of an AI strategy. So as mentioned to me, an AI strategy should be grounded in these foundational perspectives and maybe you have a couple more to add to those. The biggest pitfall that I see is that an AI strategy becomes a sum of use cases who get crowdsourced from existing departments, existing functions within the business, and then get pulled together in an AI transformation function. And you get a roadmap of use cases, which in my mind is not an AI strategy but an element of an AI strategy. An AI strategy for me is a coherent blueprint of how to win in an AI native world. That requires to first form a vision of what that AI native world will be, how that will change your industry, how you want to win in that industry, what your competitive advantage will be, and then work backwards across those three waves of transformation which we talked about, getting to your roadmap and your actual use cases rather than the other way around. So this is the full blueprint of all the elements that we typically look at in terms of AI strategy, starting from the top at how that market is going to change the core elements of the AI strategy itself, the roadmap, and then the AI enablers which get you to that human AI system which might get you to competitive advantage. So but first we will need to think about what will change, which is typically what you tend to do. There's a new technology, so we're going to imagine what will change. But at least equally important from a corporate strategy perspective is all the things that that won't change. And in that sense I've always appreciated Chef Bezos point of view at Amazon, who banks his strategy on things that won't change. Like the fact that in E commerce in a decade from now, people will still want their goods to be cheaper and delivered faster to them. So then the question becomes how can I deploy the new technology to make that happen rather than just being focused on what will happen change. So thinking through both elements, both sides of the coin, to form your strategy. And here's an example that I won't get into. If we think about what will change. What's a good mental model for me is to think beyond the first order possibilities and also think about the second order possibilities. And I'll just give one or two examples on this visual. So in the center you start from what are the AI capabilities right now? The fact that it can generate things instantly in seconds, that it's available 24 7, that it's multi language, has an IQ of 120, etc. Etc. Which then gets you to imagine what first order possibilities are. For example, the fact that OpenAI's operator right now and other agents that are in development start being able to buy products for us or recommend products for us might actually mean that brand loyalty will be less important in the future. It's a potential scenario. Whereas right now humans buy products based on a certain brand loyalty for a specific brand. We might get to more algorithmic buying where AI assistants buy products for us based on what we need right now in terms of like what hydration that I need right now or how stressed I am not tied maybe to specific brands. So those second order consequences or possibilities are typically actually the most disruptive ones and the most interesting ones to keep on your radar. So there's a couple more examples here and we will share all of the slides that you can read through. But just wanted to take you through the logic or the model of thinking through first and second order possibilities. So if we've imagined what will change, what won't change, then we can start building our future back AI strategy and also inside or outside in strategy setting a clear ambition that is linked to your corporate strategy defining how you're going to create competitive advantage. And I think the most interesting lens to look at that is how it can reinforce your existing advantages rather than potentially creating new ones. Sticking with the Amazon example as just shared and then stating your key objectives, I would like to invite you to aim for the right balance between margin optimization and cost efficiencies and growth oriented KPIs and objectives from that lens. We would then start building an AI roadmap across the three waves as mentioned earlier in the foundational perspectives. Wave 1 efficiencies then getting to better quality and then getting to entirely new systems. So in terms of marketing, just a very simple example, we've already talked through that example of video generation, so just adding one more for product development as a function. So in wave one you could look at deploying AI to analyze customer insights faster. In wave two you could deploy it to actually getting to better customer insights that maybe competitors won't have. In wave three you would rethink the whole system and you might be able to get to an autonomous product engine that continuously tests and deploys new products or new product features in a digital context, which would skip a traditional human insights function altogether. So what I'm advocating for is for each of those functions to think through what wave three transformation might look like and then to sense check if the wave one efficiencies that you're focused on Right now would still fit within that model, which if they do, then all fine and which they don't, then maybe that's not something to focus on right now. Then just briefly mentioning that underlying layer of AI enablers which are crucial as mentioned before, especially the operating model and obviously the talent and capabilities you bring alongside the technology which gets you to the full blueprint of the elements of an AI strategy. As mentioned, I wanted to end with one concluding note on how to lead this transformation as a leader into your organization. And I think the key challenge will be to match vastly different clock speeds. And to me that's an interesting perspective to look at things where AI is capable to generate and create new things in seconds and minutes. People are typically able to do stuff in hours and days, but organizational structures will be able to do things in only weeks or months. And Amdal's law specifies that it's always the slowest component in any system that really defines how much performance improvements you can get. So when you deploy a new AI system that can do stuff in seconds or minutes, but your org structure still works in cycles of months, then the AI system won't get you as much. So it's really about the combination of the the three things. And so our rule of thumb is that for every dollar you invest in technology, to consider investing $2 in people and change to really unlock the value from the new technology. So want to wrap up there and leave some time for one or two questions with three key takeaways. AI transformation in my mind will happen in three waves. And in terms of AI strategy, it's good to think about those three waves as and when you build your roadmap. AI in itself, although popular belief will not lead to a competitive advantage, it can only be in a kind of unique combination of AI people and the operating model that you can combine with that. And as mentioned as well, there's a lot of focus on what AI will change, but I'm advocating for setting a strategy on what will change, but also equally important on what won't change. So I'll stop here and turn it over to you, Kevin, to see if there are any burning questions.

There are actually three. Let's see if we can get through them. Thanks Phil, for your talk. One of the things that the audience brought up was, you know, when you say the cost of AI is approaching to zero, are we really looking at societal costs like we we learned about the impact on the environment, for example. So how can companies ensure that their AI strategies are not only like profit driven, but also contribute positively to social well being.

That's a great question. And I guess that plays overall into any type of investment discussion. So I wouldn't say that it's specific to AI. And obviously right now you see investments into renewable energy, or at least cleaner energy alongside the big AI investment. So I think there's going to be a clear need obviously for clean energy to run all of the AI systems. And I would argue that is the most important. Alongside that, obviously you should have an AI governance, AI ethics charter, etc. Which we talked about in the previous session. So I would argue sustainability and ethics to me are the key standout points to focus on. And they were also very much part of the enablers in our AI strategy blueprint. So we would always work through those elements as and when we said an.

AI strategy that makes sense to just general business strategy in general always has those risks you have to manage. There's another question which is interesting. So do small and medium companies have an advantage when it comes to AI strategy of not being kind of burdened by red tape that big companies are subject to? What have you been seeing at boi? What have we been seeing?

Yeah, another great question. I think there's lots to learn from digital transformation, where indeed there were a lot of new entrants that really captured that Wave three transformation. So my prediction is that there will be again a lot of new entrants, smaller companies that catch the wave 3 transformations in AI as well.

Okay, great. Well, thanks again, Phil for your talk and I invite the audience to stay here if you want to stay on the main stage or navigate to another room.

Thank you, Kevin. See you later.