Webinar: How to transform your business towards an AI-first operating model

Okay, let's, let's kick off. I would say we'll need the time, use the time and I don't want to take your Q and A, even though that's always a challenge, obviously. So what are we going to talk about? We're going to talk about transformation. We're going to talk about an AI first operating model. And yeah, we have been talking a lot about innovation, about using AI in innovation, have been talking about the tech. But there is a big part of the autonomous transformation, as we call it, is people and is also the culture and how it works together, the processes, the governance, the business model and so forth. So I would say consider this a first grab at how you can really think on a business level rather than just, I would say front end innovation. What I'll do is also I'll. As many of you might know, we had the summit, our autonomous innovation summit, two weeks ago, which was a tremendous success again, so with more than 10,000 people joining, more than 3,000 companies again, all the big tech players from Google to Nvidia to Microsoft joining and then mit, Harvard and some of our clients as well, big global companies and then also the startups, the tools that we work with, for example, Outset or builder AI in terms of seniority, it's pretty widespread. So from CEOs to consultants to students to directors here with an autonomous operating model, I assume it's most relevant for the CEOs because they have the biggest mandate. At the same time, I very much realize that not all of you can be CEOs, so I really want to talk to you, assuming that there will be NCEOs and mentors and directors and so forth in the room. So yeah, let's try to be relevant for all of you. I will also come back to interesting elements that we heard during the summit. So if you saw all the talks, which would be pretty impressive, then you might see some things coming back. But of course it's all stitched together so some repetition does never hurt. A slide that I will end with, but that I will also start with is there is three things that I think regardless of your seniority, you can always, you should take away and should do is first of all really make sure that you have a strong perspective on what the autonomous world will be and how you want to play a role in there as a person and as a company. Then the second one is seeking holistic scope. So you have your own role, you have your own department, your own function, and you tend to look into that smaller, I would say function but it really gets really interesting when you think about the entire operating model, thinking about end sales and purchasing, NRD and so forth. And then once you find that holistic scope, either you have it yourself, you have that mandate or you try to find someone that has, be it a business unit leader or a brand or so forth. Then you try to demonstrate that holistic value. So these three things, try to have this in the back of your mind while listening to this presentation. I'm one of the or Arnie for most internationals. I'm one of the partners of Board of Innovation. I'm a product designer by education. Very quickly went into business. I have been pretty geeky about for example how it's made. I think I watched all the 32 seasons so I'm pretty obsessed by it over the years. And I'm very, very much a people person. I grew with boy, I grew boi and yeah, I've been part of many transformations as well with our clients. Probably good to give the word to Jeff as well who is with me.

Yes, hi everyone, Jeff Gibbons here. I lead the US business at by based out of New York. Delighted to be here with you and Ana today. The work that I've been doing. We'll talk about some of the examples of how we're working with clients to actually transform to this AI native operating model. And excited to share some insights with you. Maybe actually to talk about that if we skip to the next slide. I think what we wanted to just start with is it's an incredible, interesting and exciting and terrifying moment. I think that we're all in, in terms of the emergence of Genai and how it's having an impact on the way that businesses are running, the way that teams are feeling about their careers and their futures. And I think maybe just start with a few of the things that we're hearing from leaders so that you know, there's a huge amount of ambition, opportunity but challenges. So you know, for a lot of larger companies they're seeing that actually the emergence of AI and using decades of data that they have is actually a really interesting opportunity for them to compete against startups. There's like a positive thing there. There's really an understanding of like how can we actually help our teams to stay within the realms of what's possible from regulatory standpoint. So there's like some ideas of how can we guide our people to make better decisions on a day to day basis. There's obviously opportunities a lot of companies are looking at. How do we actually decrease our dependence on agencies and consultancies. If we can automate knowledge work, couldn't we actually be, couldn't we amplify our people and make them less reliant on using consultancies? But I think there's actually some real challenges that a lot of companies face in actually pursuing this. So how can we actually find the right people? How do we upscale our people? How do we change the way that our people work? How are they going to feel about all of this? And then also how can we make sure that we're doing this in a way that's positive, free from bias, and actually makes the world a better place and improves sustainability? So these are some of the types of challenges that we see from leaders that we talk to.

And these are all concrete requests that we're getting on leaders thinking like, yeah, how are we going to do this? Can you help us? Some of these we have been actually collaborating with some others are doing it on their own. Maybe a bit of a couple of words on the last probably decade with Board of Innovation. I think first of all, we have always believed in the potential of Umont Plus Machine. In 2016, actually Phil, our founder and CEO right now I pretty much see him like giving an all teams presentation saying okay, like we're going to be like 50% human, 50% machine, super inspiring. And then it settles in saying like, yeah, how exactly are we going to do that? And yeah, technology wasn't ready at all. And to my surprise, even our team of about 50 people back in the day, yeah, they weren't really ready either. And you would expect or I would expect an innovation agency, design agency that hires for very open minded, change minded, resilient people to be ready. But it's really difficult. So we are in general geared against changing. So let's take that as a given and let's see how can we bring our people to this journey then? Yeah, of course as a company bringing this to others, we are eating our own dog foods for the last from 2016 at least. So we started in 2009. We started as a business model innovation agency. Then we added design to that and we added a couple of other things to that. But we were an agency like many others. Then when AI matured and then especially of course the end of 2022 when OpenAI launched GPT. Yeah we really became AI powered. We really collaborated with AI again like many others. But then later we really grasped the potential of like if we really, really help companies to reinvent themselves, then we can work towards autonomous growth. Growth has always been Something we offered and it gets really exciting. And this is pioneering work. We see it, we're discovering it together with our clients and with all the great companies and speakers and people that attended the summit as well. Obviously, maturing tech was one thing. On the other end, a rough macroeconomic context helped as a hand or two probably don't have to tell that the last two years weren't like great for agencies apart from those that had like 100 AI engineers, I guess in their workforce, especially IDEO, I think was a big shock for the markets when two years ago they actually let go of about 35% of their people. So that was a big one. And that really pushed us to say, okay, like we wanted to do this, like let's really double down and go all in. And that, yeah, that proves to be the right move. So then one of the frameworks that you can take away from this session is what we call the reinvention blueprint, where we talk about the strategy, where you really have to define, okay, how can you become a leader in the autonomous world? Where you think about, okay, how are we going to improve our growth or efficiency? Competitive advantage. And actually competitive advantage is a very important one here then. Yeah, the engine, which is more the product element where you say, you might recognize the built, measured, learn in here. So where you learn in a very always on way, where you then generate new propositions, build them, launch them and simulate continuously and always on. So that is a very important element. And then in order to scale this in an organization, you have to have leadership on board, you need to have your people on board, your culture needs to breed this. Otherwise you're always going to like fight against your own people, which is difficult. And yeah, for you and for them, then the business model, technology, net positive impact will touch upon these elements. But actually that bottom layer is what, what I would like to talk to you about most right now. But I will touch upon strategy and engine as well. Yeah, so this was a reminder slide for me because I thought I was going to forget. Probably relevant for you also to know like what we have done ourselves. Yeah. So first of all, I would say we have brought the vision internally. Now, eight years ago on saying half human, half machine, we didn't put it out there right away, but we definitely started experimenting. We've had quite some failed experiments in the beginning with, I won't go too much in depth, but at least everyone was experimenting with it. Yeah. Then we started like hiring, like evolving our people to more product owner type of profiles because we Started really building instead of like facilitating the entire process. We have been in terms of culture also like we have always shared our tools. We were one of the first Companies that had AI GPT tools on our website January 2023 on innovation to create Personas, to create future scenario and so forth, which was with GPT API behind it and maybe one on net positive impact there. Everyone always says about sustainability it's really important and our clients also say that. But it really comes to it, it's painful to see that it becomes a nice to have. Then nobody will. Will. Will say that. That's true. But we see that. So right now when we actually code a product, an engine, then we can really like build it in from the very beginning. So if there is a thousand propositions and we say like, yeah, we're going to present you 100 and it's these hundred that are also like way better for the planet, then we can. We really have a better story to tell and we also feel better about what we work on. Maybe. Jeff, any others that you feel are relevant to share?

Yeah, I think the key point here is really how do you sort of have the. How do you think about how all of these different pieces work together? And I think that's really where the competitive advantage will come for a company in transforming is not just that you have the technology or the leadership or the right data. Technology is like, how do all those things work in sync with one another? That's really where we see that companies will develop that competitive moat. And that's really kind of our focus as a company for our clients is how do we get all those pieces humming together?

And then for us, maybe a last one about like eating our own dog food. The like, the whole summit is also something we created for, I would say us and the community. So like two reasons. Like, first of all, you need to learn from the best. And bringing the best together is obviously good for. For us, for our own team and like upskilling them. And at the same time, the more people that are aware and follow the vision that we have. Yeah, the more, yeah, the more momentum that will be created in the market and so forth. So there we really like to give, create market with it. And then. Yeah, well we have always received back from community, so that's great. Then for those that weren't at the summit. So the world will become more autonomous. I've been speaking by autonomous quite a bit. So what exactly does that mean? Let's first look at this AI faked TV episode from South Park. And I just cut out a minute to give to make an example here.

Guys, have you heard about the screen actors strike? Yeah, it's all over the news. That sucks. Does that mean there won't be any new shows coming out? Exactly, Stan. And that's where my business idea comes in. I'm glad you asked, Kenny. It's called qupy, a deep fake streaming service. Wait, what? Deep fake? Cartman, that's illegal and unethical. No, no, no, it's genius. We take existing shows and movies and just replace the actors with deep fake versions of other famous people. Like imagine Game of Thrones but with Danny DeVito as Daenerys. That's ridiculous, Cartman. Is it ridiculous or is it so ridiculous?

You get the point. So yeah, this is obviously, yeah, this is South Park. It's cartoon. It's easier than real movies. But AI is progressing and this is actually like a parody on Showrunner launching now about two weeks ago, and they claim to really create fully AI generated TV shows and that everyone, they will make it available for everyone to do so. Yeah, you can imagine that it really, that is really going to disrupt how the entire movie industry and like serial industry works because yeah, the entire value creation and yeah, will. Will be turned. Yeah, will be turned around and we'll see how it, how it evolves. But it's definitely interesting to see. Yeah. Then someone I was like speaking like the founder of. Of wizard was also speaking on the summit. So, so what you see here is a demo of this tool where you actually can create, let's say basic UI for apps, website and so forth, and you can really can have it generated and then you can interact with it and then you can also export that to figma. So like, why is this super cool? It is cool because someone that doesn't have AI UI knowledge can really do this and then can also collaborate with designers to say, okay, like, like start from this. And then it's a way better brief. It's also usable for the complex tasks. And this also means that designers can focus on the things that are complex that they learn from, that are not AI possible to generate with AI yet, which at least our designers are very happy with. Because a lot of the work of designers or any other generated, like, like designers or people that generate stuff goes to a repetitive work which is, which is not great for them. Yeah, then fashion industry.

And I think if you actually view and if you extrapolate this story forward, you actually see a complete transformation of how a digital product company runs because obviously we're seeing here that you could create a screen like, you know, rudimentary version of it created pretty, pretty quickly automatically. Fast forward two, three years from now, you could imagine actually that a digital product is doing this by itself. And so actually what does, you know, so imagine I log into an app or something. It would actually be creating screens based on my interactions each time I'm seeing something. So the work that a design team or a product team will be doing in that company would not be actually designing individual screens, but actually designing the overall system and decision making logic for how screens are created in real time. And so their work is going to entirely shift in the next two or three years and really it becomes more about how do they create decision making systems rather than designing and figuring out where the button goes.

Yeah, yeah, exactly, exactly. And then some other examples is like Shein Fast fashion is like ultra Fast fashion right now. Shein launches a thousand new products a day and it's controversial and everything, but sustainable. We'll show you an example where we actually have created a better version with a very big retailer. And yeah, then a lot of the work is focused on B2C retail, but there is a lot to do as well when you look at material science and B2B in industrial. And an example here is Google DeepMind, like researchers working together with Google, Google DeepMind and they actually created 32 million new materials and brought that back to 80. Yeah, 80 potential or sorry, 23 potential materials or crystals that could actually improve batteries. And they have done that in 80, in 80 hours. And then these 32 can be tested in the lab. So that is, yeah, normally it would take not years, but like a couple thousand years to come up with all that knowledge. So this is super interesting and this is the very, very first step of work that can be done in material science but also in health care and so forth. So about 18 months ago we shared this slide in another visual identity. But we said like almost everything will be AI generated and this is very much becoming, becoming a real. And it's in our opinion, very, very exciting. And it is also, it is because there's a lot of benefits to innovating in a more autonomous way because you can get to better products if you design the engine well. So for example, if you say, okay, sustainability is key and I also only want to see products that are sustainable, you'll be way faster. So there on the fashion example, you're going from months to minutes, it's going to be more cost efficient. It is always on and always on what does it mean? It's essentially you give a machine access to different APIs, which, for example, can also be social listening API, where that social listening always looks at, okay, what is happening, what is trending, and then acts upon that. And if you do this well and you really create an operating model, it can be a competitive edge. It can really competitive advantage. An important element is here that everyone has access to the APIs of OpenAI, that everyone has access to the tech. So if you don't look at a bigger operating system, if you don't look at the proprietary data that you have, then you'll likely not get any anywhere near a competitive advantage as a company. Yeah, so very exciting. At the same time, a lot of questions are prompted. I'm not talking about GPT here. Just like a lot of questions arise. There is the reality that this is like complex. Not many people get it. It's going so fast that we don't even like always get the full complexity of it. There's a lot of fear, there is a lot of uncertainty related to AI. You could say, like, yeah, governments will help us and they are definitely trying, but we definitely don't expect them to lead the way here. So we'll likely phase a long wait if we wait for their clarity and frameworks. In the same time, there is a lot of people that are fearing to lose their jobs because they're not aware or they're very aware and they see what it can do. So here, these are realities and AI is here to stay. So let's embrace it. Like, let's embrace the progress. Let's, let's kick over our journey today. And that's why we're here. We really want this talk to be. Yeah, a kickoff for you. An important element here is like, we don't believe in a fully autonomous setup. It might be possible, but that's not what we want to build for. So in every industrial revolution or any big progress up to now, humans have always been the hero in the story. And you might design it in here. And you could argue on this one because AI is smart and many smart people argue about this. But I would say our version of the future is with humans as the heroes. And what does that mean if you think about a company? So we're going to manage autonomous organizations by amplifying the power of your people. So yeah, let's assume that you get your humans, your colleagues, your experts, you get them on board, they get it. That they're going to be orchestrators, that they're going to Be editing rather than creating. The creation is done by machines, agents. So they're like. Agents is also an important word. The future of AI, generative AI is really with agents. They can really do the execution at scale, the things that we don't like to do. And on top of that also the things that weren't possible because it just took too much time thinking about the Google Beatmind and the 32 million new materials. And then very important, each company will have different frameworks, different governance. It might be, as I would say, simplistic as just templates that you have yourself or specific strategies that you have to say, okay, we actually want the people are experts and the agents to collaborate in this way towards these objectives. And that is going to be different for each company. I mentioned two important things here already, so this is actually better for people, we feel. Jeff, any examples on collaborations that you want to share?

Yeah, yeah, I'd say like recently we've been working with a fashion retailer looking at how do we actually build an engine that actually takes information from the external world, from TikTok, from runways, from all sorts of things in the external world, use that to generate cultural trends and designs and design inspiration. And so the work that's actually being done in it is often the work that's done by a design team at a, you know, the retailer. And the interesting thing that's been surprising in that work is actually the level of pull from the design team. So we actually had to be holding back access and usage of this because there's still some things we need to figure out to make sure it's safe to use and they can't use it for producing things yet. But the pull from the design team has actually been very strong because they really see that there are many parts of their job that they find a bit laborious and they see this as a way to actually free up their time to use more of this, to use more of their time not to do those things. And so I think obviously the, you know, the typical story is around how AI is going to take people jobs. We haven't seen that in all instances. We've also seen a lot of pull from people to help them and amplify the output of what they can actually achieve in a given day.

Thanks. So here you probably have to do it, show them in order to actually get them convinced or get yourself convinced.

Yeah, it's harder when it's conceptual, but when people actually have something they can use that they can do something that usually would take them hours, that takes Them minutes, that's when they actually get excited and see how it could actually apply to their day to day life.

Exactly. So with this autonomous future, which companies will really thrive and I'm not talking about Nvidia here, it's those again that built that AI native operating model that is a true industry game changer. So it is important that it's an industry game changer. One example also talked on the summit is notco. So I'm really impressed with what they've been doing. So they started seven years ago, it's a company that by now is worth about 1.5 billion. And they said okay, we feel that the world needs to cool down so to say and we want to help with the end and we're going to replace all the animal. Yeah, all the animal into planet based ingredients in products. And that is, that is what we want to do. And yeah, their mission is to really like make it accessible for everyone, get it into every meal. And they have built their own AI. So they have built their own AI where you have like a lot of different ingredients and it makes yeah, formulations for new food. Sometimes it like puts together like pretty interesting things. If for example, like recently, yeah. Came out with a, with a chocolate actually that doesn't have cacao, which is interesting especially knowing that in the last year ago or in the beginning of this year the cacao prices like went up or like times three or something like that. And interesting here is also. So they started building this with the intention that they were going to sell it to chefs of big CPG companies, to product development, to R and D of big product development, to R and D of big CPG companies. But yeah, what they learned was that nobody cared about their AI. They just wanted to see okay, which products can you make with it. So they struggled quite a bit but their AI was great. So what did they do? Started making their own products. That's also when they started when they needed funding and so forth. And for example the nut milk was done in collaboration in Starbucks, with Starbucks and they have created about a hundred different products, brought them to market in 4 years time which is pretty incredible. Like knowing from the perspective of food and beverage companies and the amount of products that they launch. They also set up a joint venture with Kraft Heinz and for example said okay, like yeah, what can we, what can we do with your Mac and cheese? Can we actually make it plant based? They could and they did. It's pretty inspiring as a company. An extra element, especially if you're, if you're The CEO of a stock listed company company. Yeah, they're actually, they're valued or their valuation multiple is the multiple of a tech company which probably every food and beverage company can dream of. And they really organized, they organized that way. So they have their, they were launched in Latin America, now they have their headquarter in San Francisco. To attract tech talent. They are having their B2C owned products angle. They have their B2B angle where they collaborate and like make products with. Yeah with food and beverage companies with the extreme craftyinz being a joint venture and it's really, really well integrated. So they have R and D supply chain production lines, marketing brands and so forth. It's really, really integrated. So we're not talking about, about an innovation function or a marketing function or something. It's really, it's really integrated. So for me this is the. Being the best example that I know of that we weren't involved in. I would say if you are interested, look at the talk of Adit Patel. If you go to the autonomous innovation summits you can still register if you didn't do it yet and then you can have a look at all the talks. I'm going to launch a quick survey and I'll give you some context here. So if you think about NotCo, there is different ways you can, you can do this. So either you say you're going to venture this, you're going to start this from scratch, you're going to build your own AI and you're going to make that change in the, in the industry. But you realize actually it's going to be too slow if you do it within our company. So we'll, we'll spin off and we'll venture something or you find a partner, you'll find or help the Notco in your industry. There will be more, more of them popping up. Or you transform your current company and evolve your current company with the majority of its people and organization towards an AI first operating model. You should be seeing the poll right now. I'm interested in getting your votes. And you don't have to answer Transform by the way, because we talked about transforming like the other two are definitely also interesting. And if you, if you want to give a bit more context to it, feel free to use the chat. See that they go up quite at the same level. So I'm interested in the wise actually behind it. I'll leave it open for. Let's see, I think about 50% of the people voted. So we have transform as the highest, then partner as the Second, and venture as the last one. It very much depends on who you're like, which company you work for or the company that you have, and also the industry that you have. So, yeah, let's definitely share and have a quick discussion about this. I'm going to close it for now and I see that not many say venture, which is indeed it's probably a difficult play, the venture one, but the other ones as well. So I'm going to close it and continue the talk. So how do you seize the opportunity? You set your strategy, build a product, and a product is either the engine or the actual products and then you drive the transformation. So in terms of strategy, in order to become more autonomous, you need to have gone from analog to digital. That's going to be the easiest. And then from digital to autonomous. Many companies are still working on their digital transformation. It's possible to skip that step. But these are usually the different steps that you take. And yeah, I would say those that are working with AI already, what we see right now is that a lot of companies are only looking at AI as a pure productivity tool. And yeah, we feel that that's a big mistake. So there you're really not capturing the full potential of AI or the autonomous potential it has if you do so. And to give you another framework is if there are different strategic options you can think about like what do you launch to market, what propositions do you have and then how do you do it? So if you think about the what's you have existing products and services and you have AI native products and services. And then when it comes to the operating model, again, you have your existing operating model with your existing channels and so forth, and then you have an AI native one which will be new for majority of people here. Yeah. What does that mean? So AI efficiencies is an existing operating model, existing products and services, then moving to AI systems. So that's not co. Then AI products is going to be mostly focused on additional services or products on your current portfolio. So it's going to be mostly about customer experience we see. And then there is AI breakthroughs. So really disrupting the market paradigm on that last one. Everyone wants to have that. But just looking back at the digital wave, the big, big market shifts haven't been captured by Inhumans. So there it's actually the Netflix, the Hellofish and the Teladocs of this world that really captured things in their market. And it is often a winner takes it all, but it doesn't necessarily need to be. I already shared notco showrunner. One that I didn't share yet is the Lancome. It's like l' Oreal's products or service that they have in their retailers. So you can actually scan your face and the AI will tell you the age of your face, like what you can do. And that way you get advice. And in that way also the people, the salespeople in their shops are enabled to give you much more advice based on technology. They have a similar version that you can use your phone for, which is a little bit more basic. But it's interesting to see that they do it per channel, both retail and for their E commerce. So there let that be an inspiration for you as well. Yes, the race is on. I would say every industry can definitely have at least one not go. So let's see how you can evolve or partner up with that company. And yeah, for you, when it comes to strategy, there is no like, you don't have to choose. Think about a portfolio. What can you do when it comes to AI efficiencies? What can you do when it comes to operating model? What can you do when it comes to the AI products and to increase your experience of either your end clients or the operators or the channels you're using now when it comes to building the products? So I would say LLMs are great and they're incredible. There is definitely in order to grasp like the full potential and to really productize the full capability. Even with the current stick, it will take years and we feel that the tech will always be ahead of us with the current speed. So let's really work on it right now. And we are really on that journey. The only way of fully, fully transforming and doing it is doing it. So it's really like reinventing how you do, how you make things, how you buy things, how you use things. So that's that middle part of the engine. And here I'm going to give the words to Jevigant to talk about a case or a client collaboration that we have.

Yeah. So this was, as we mentioned earlier, working with a fashion retailer looking at how do we actually first of all build an engine that can actually create designs and figure out assortments and collections, but then also, what does that mean for the way they actually work? So we actually started out building a proof of concept of the engine and then partnering with their design team to figure out how do they actually integrate into their work. And so from the development of that engine, we found out that the design process could go from months to minutes in terms of actually creating a usable design. We actually validated the outputs of this and we found that the outputs in terms of the garments were testing better with consumers as it related to purchase intent, as it related to the trendiness of those designs, and also as it related to the visual cohesion across the line when they compared them to their existing best sellers in the market. So not just like what they sold, but the existing best sellers of their existing line. And so that, that really then became a point at which it said, at which the company said, well, okay, this, this works, this is great. And so what does that mean? And how do we use this and how can we make this safe and actually integrate it into our ways of working? What does that mean for us? So we moved and I think really the point was actually it's not just about having like a design tool, but actually how can you use the tool to actually rethink the model of how the overall business works. So if you have a tool that can automate design, how can you actually integrate that into the way that products are purchased from suppliers? So how can you actually reinvent the way that suppliers will bid on a design and whether they're going to produce it? What's the pricing model around that? And so it's not just about like a fancy tool, it's about how do you use that capability to rethink the entire model to make their brands more relevant and also to do things like reduce waste. And so we looked at from the initial kind of design piece kind of in the bottom right, how do we actually integrate other pieces of how do you actually figure out then using the same data, using the same kind of engine? Well, if you have that product, where should we sell it, how can we produce it, which ones should we produce? And actually reducing the amount of waste in the industry. So about 20% of all garments produced are currently thrown away or burned. Just because you can't predict how to, how to actually do something, you know, what's actually going to sell? And so we actually have now looked at how they actually have a new go to market model where they actually launch an image of a product before they've even made it. So you're actually introducing a more real time demand prediction function into the business in terms of, to address the question from Jacob in the chat, we actually worked with the company to build this engine. It actually relies on seven different LLMs within their environments to actually perform different roles. So there are four different multimodal LLMs that are used for different parts of the process and Then also three image generation as well. So lots of different lms. And it's kind of a flexible architecture so that as some new LLM comes out, you can kind of switch them in and out rather than being reliant on one single LLM. And so as I mentioned, really the, the importance of actually how they change their business is not just about you can do your design quickly, but then you can actually take that automated design process and rethink the way you launch something. So you launch a picture before you even made something. And then you can actually de risk the prediction and reduce waste in the industry. And then that actually completely changes what the entire model for the business looks like. And you don't have to be relying on seasonal collections, for example. You could launch things all the time. And so as I mentioned before, so the designs actually outcompete the current bestsellers. And the decision, the time to decision around a product now goes from about six months to a few minutes. And they were rated better by consumers on three different factors, cohesion, purchase intent and trendiness. And then also the actual designs were rated. We actually tested them with external designers not within the company for sort of a non biased test. And 75% of those designers rated the actual design files themselves, the sort of blinded outputs of those as very good or excellent. And so it's once you have that kind of level of quality, you can then introduce automation into lots of different parts of the process. And so really the way that this works is actually through a variety of different agents, as Ana mentioned earlier, working to solve problems together. So it's not just that you have like one LLM that's producing an image of a product. You actually have different LLMs and different agents relying on different LLMs working together to solve a problem. So you'd imagine you'd have like a, like a creative director agent that is figuring out, based on what I see in the market, around a cultural trend, what is a design that actually fits this brand. You then have a merchant agent that's figuring out, well, where would this go? Do we need this? Is this something that actually fits within our collection? You have a procurement agent that's figuring out what's the most cost effective way to produce this. Could we produce fewer of these and still meet the same kind of demand? Like make eight of these rather than 12 of these things. And then obviously like a consumer agency, have like a synthetic consumer agent that is telling you which ones, you know, which ones consumers would like the most. And so you can imagine the system is Basically generating hundreds of designs and then working with these different agents basically conversing with one another to tell a marketer or a merchant here's the ones you should make and here's why they're the right ones to make because they're the most efficient and it's the fewest products to meet the most demand for the consumer. And yes, that would be the AI agents. Yes, sorry, question. Okay, back to Ananda.

Yeah, and I would say that so where we are right now is like we are up to the tech pack. So they're like in order to like read a production file. And for me also something that is interesting is that the always on also means for example like having the material library with actual costs linked to the generation of new apparel. That's usually different people looking at it and also sequential. So that's really cool. So then when it comes to transformation or getting at it, the key to success is as I mentioned, it's not the technology that everyone has access to but it's really designed on that operating model to unlock the volume. And to give you an example like March 202023 paper came out, Bloomberg spending more than 10 million on trading their own Bloomberg GPT on all their data to then realize less than six months later the GPT4 was better than doing it. So they're like by now the companies that are trying to create their own LLM. I haven't seen many anymore. It's really going to be like economists couple of LLMs. There is now about like six that are. Yeah that are great right now and they're good at some like there are some variations on what use cases they're best for. But I would say like we don't, we never advise anymore on creating your own LLM. And then when it comes to the tick, the tick will become cheaper. But really thinking about the money you invest in tick. Yeah we advise on like investing like double of it in people and change. Yeah you don't want to start over like in humans big companies. Yeah they have a lot of assets that's really leverage these assets on top of tech because if you don't startups will be on par with you and then it comes transformation. So most of the companies will focus on function. So function for example saying like yeah, how can we create legal contracts? For example that's in order to say okay, like the function of legal. Yeah you're going to replace that while you really have to look at okay like how can you create that AI native operating model Think not go that is an industry game changer where a company as big as Kraft Heinz says like we can't do this ourselves, let's partner up because it's going to cost us too much to do this ourselves. And when I say like Kraft Heinz or Unilever or Nestle, like we're talking in terms of R and D budgets, we're like talking like 300 million. A billion, two billion. I heard as well, not calling out names, but it's a lot of money that they have and still they decide not to. So looking at the enablers. So in terms of leadership, I'll share by the way, this deck. I know that we're like rushing a bit and I want to keep like a couple minutes for a couple of questions as well. So I'll be sharing this deck. In terms of leadership. It's really about like cultivating like experiments. Yeah. Investing, learning being an example. So here, as I mentioned before, make sure you have a vision that you share your vision. Like what is the autonomous world going to look like? How are you going to benefit? Like what is your role going to be than seeking that holistic scope here I would say it might be very relevant to work with an external IT environment which might be your own sandbox that you set up or an agency that you work with because otherwise you might have this catch all policies. There is still a lot of companies that just say you can't work with AI because we haven't figured it out yet, which is obviously a shame. And then we also believe that we are hiring product owners. We are moving like upskilling people towards being product owners in our own company. We feel like all leaders need to become product owners. And if you feel like, if you want to challenge that statement. Yeah, well here is a decision tree. So if you feel like there is a lot of repeatable workflows in your business or if you have. Yeah. If you're responsible for creating competitive advantage or if you are actually responsible for predicting the future and having hypothesis when working, then either one of these three, then probably you need to be a product owner. If not, then you have to be. But that would be surprising to us then when it comes to talenting culture here, I'm just giving a couple of examples. There is many things you can do but here the one is just doing it like demonstrate the value of it. There is going to be skeptics always. So let's not give them the chance of just saying like calling your bluff. Then maximize your exposure even if it's just starting yourself the reason that you're here is probably talking to the wrong people here. Um, but yeah, for example, for me about, about eight or nine months ago, like I was going through a personal coach myself and what I did, I was like, all my notes I've put in a custom GPT and now I that, that AI is, is my coach or at least the moment that I'm not with my coach. So there, that's not even business business related, that's just personal related. But you can really like just work with it, try with it. And custom GPTs is, is I would say the most accessible for everyone. And then you can do a lot, but you can also start small. So for example, if you think about synthetic testing and a lot of them slide again. But I would say the least complex is custom GPTs and that's because they made it super accessible to for example add some data in there at a certain prompt or format in there and then you can collaborate and talk to it. And the more data you put in, the better it gets. And then you can work your way down the work towards the super digital twins of either buildings if you want, saying really you're going to create an entire digital twin that you can collaborate with and so forth. You could have entire governments in there that you can simulate things with. Then in terms of processes and business model, I'll skip through this a little bit. But partnerships is key. Understanding which battles you want to fight. So understand where the biggest benefit is for you as a company. And then a principle that I very much like is often risk and compliance. They're seen as the bad guys in companies. We really don't believe in that. What you just need to do is you need to start collaboration on the principle of maximum liability. So there it's really about like how can you decrease the maximal risk or the maximal liability which you can do in a sandbox, but you can also do by saying like, for example, like you work like off brand or things like that. So you need to have that conversation because they're the true expert on limiting risk. So maybe you can have use cases that don't have confidential data for example, or you can work with an external agency again like have them take the risk and then it's on them to do so. Then in terms of data and technology, first of all, agents, APIs, data, it's all very API driven. So if your IT architecture is not API driven. Yeah please make sure you have that. Otherwise it's not going to work. Then think about which data sets can really Give you a competitive advantage. Because for example, for NotCo, they can use it for their own products, they can sell it to their competitors, so to say. And they can also go like in the entire ecosystem. They can also work with ingredient providers for example who really like. It's a huge market, yeah, 30, 40 billion globally. And they can also use their data, put it in there in that way. I work with it and then experiment with multi agent system. I would like to give you a little bit of an overview because it seems super complex, but you can actually already on your own with a very small budget call out. They could say like a hundred bucks a month. You can build a multi agent model because what do you need? You need a knowledge base. And the more knowledge or data you put in there, the better it gets. But you can also put a little data in there. Then you need specific frameworks which in the most minimal setup is just a prompt. But you can of course like have much more there. These two together, you're feeding different agents. And these agents are, yeah, I would say AI, yeah, service agents that really think very narrow because you design them that way. You want them to be that merchant or you want them to be that designer. You want them to have something in particular and they need a particular skill set. So you link one particular LLM to them, might be OpenAI, might be Gemini, might be another one. Based on exactly what they can do. You define also what outputs they need to give. Is it text, is it video, is it voice and so forth. And then there is tools where you have, for example like Cassidy or Dust, if you need a European one that actually make them talk to each other. So you can like put them all in the system and you can say okay, you can have them talk to each other. And then super important, please keep this human feedback loop in here. You have people that collaborate with these agents that give feedback that make the outputs better. So this is super key and I would say in all transparency, without that human in the loop, it's also not good enough yet or what we see. So you really need that human in there. And then as a last one, net positive impact, what we're talking about here is for example like AI is, is indeed like it can generalize, it can really be unethical. So they're like be aware of that bias it can have. Then what I mentioned in sustainability, you can really code it into your engine. Say okay, like you have to have beat Nutri score or carbon footprint or something like that very early in the process. While usually it comes later in the process. Yeah. And then just keep on remembering like people are your heroes. It's always going to be. Yeah, it's always going to be good for you if you help people transitioning, upskilling. I probably have an entire talk about that. So. Yeah, help them to help you. So concluding here, we're going from a digital mindset to an autonomous mindset with humans as creators that become editors where you have like a product led operating model and product led. That's from a leadership perspective because it will obviously AI will be AI first and then way more integrated than it was up to now. And to go back to these different steps. So here, regardless of the seniority that you have, make sure you define your strong perspective, SQListic scope and demonstrate the value. And we can obviously help you with all three of these. But let's just start yourself and that's really key. And I promise you a couple of minutes. I tried 20 minutes, but it's two. So let's see if there is any key questions that you want to ask. And I'll take away my presentation so you can see us. Jeff, did you see any questions or pins? You're muted, Jeff.

Sorry, I was getting unmuted. Yeah, I think there was a. There was a question around sort of how do you choose the right LLMs? I think for the right tasks, because we talked about the example but we had seven different LLMs for one engine. Yeah. I think the point is actually partly. So there's some advice in the chat about using something like Mind Studio. Partly the goal is actually just to try them out and see which works. Obviously it would be simpler to do fewer LLMs. Typically we sort of work with something like OpenAI or Gemini and then see where it doesn't work. And then that's where you would look to bring in another LLM. And then advice for small firms. I would say I think the first thing is just sort of actually just get your own compliant environment. That way you know that you can protect your data confidentially and make sure that you're not losing your data to model training. So that's like the first thing. And then really just think about how could you start to make some basic changes in productivity and start to see how you can create some internal belief within the team around the benefit of those before you sort of start to make massive changes in a small environment.

Maybe like Fadim. You're asking like talking about a lot about product driven perspective. What about for example food service driven companies like retailers or for instance. Yeah, I would say there, there's a lot of people working there. So really thinking about experience and convenience. In the end, the more repetitive tasks that you can take away from the people working there. Yeah. The more they can actually invest in human relationship that they have with clients, is what I would say. So if you have that framework, really thinking about the AI first. Yeah. AI first type of products. And I'm by the way, happy to say a couple more minutes for those that want to still want to ask a couple of questions. So then. See what is the last one. Q and A. Sorry. Yeah, so a little bit more about the. Yeah. Maximum liability for risk management. Framing a lot of words in a sentence. Yeah. So I like often the principle that risk managers use is they want to eliminate risk and that's their job and they're good at it. And from the moment there is a potential risk, they extrapolate that risk and then it becomes like for example, saying like, yeah, you want to test something in market and you do it on brand. Then they say like, yeah, what if there is. Something is wrong on the packaging or something like that and somebody like shares a. Shares a claim on that and it becomes so big and then like before you know it, like your company is bankrupt. While you could potentially also say like yeah, we just want to share it to. Yeah. To 50 clients in an. Yeah. In. In a shielded. Shielded environment, they're all going to like sign an NDA that they know or they come to like share that they know that it's new products, that they can't put any claims towards us. And in that way you, you make sure that they. Yeah. That you have risk on. Yeah. On board rather than. Rather than against you. And if that doesn't work, just push the risk to others. And I, probably not many are going to be happy that I say this, but like we often, we often do that because that's the only way for. Yeah. For us to work with some companies to just say like, yeah, we'll just bear the risk.

Yeah.

Yeah. And then in terms of like, who can usually drive it? So I started with that. So it's really about the holistic scope. So obviously C Suite is most obvious, but you can also think about like business unit. You can think about brands, you can think about a category, you can think about if you have even just one restaurant can be a team level. Ideally you have a multidisciplinary team in that case. So. And if that all doesn't work, then just start proving things on the, on a functional level. Even Though on the long term, of course, it's best to seek that holistic scope.

Any. Maybe, maybe one last one. So any tips on proving that an LLM actually works? Would you advise a B testing? Yes, I think that's one of the main things that we found helpful is to actually do some AB testing of what is the sort of traditional human led, whatever you call it, approach and what's the impact and the change and sort of decide on one of the right metrics, whether it's time, whether it's quality, whether it's repeatability, whatever it might be for that task and very quickly develop some kind of proof of concept that you can develop an AB test around. We found that that helps two things. One is actually just the general confidence that it's worth moving forward with. But actually that also helps to create the business case around moving beyond a proof of concept. And typically we find, you know, it's hard to get organizational buy in around the idea of transforming your organization if you haven't even understood that this thing is good in the first place. So having that kind of a B test is a critical step in actually creating some kind of value case around the idf. Transform. Transforming to an AI native operating model.

Yes. Okay. Yeah, there is definitely a lot of questions left and we can continue here as well. If you want to have a look at the talks that were given at the summit, you can go to the Board of Innovation and then tune in I believe, and then you will be able to like find the Autonomous Innovation Summit or just Google it register and then you have access to all the talks. And there is, there is more than 50. There is also actually more than 100 because it's the second summit. It's super, super, super interesting to look at. And if you're not a tech person, like the custom GPT. Olga, seems like I see your question popping in. I'm not a tech person either. And you don't need to be a tech person to use custom GPTs. And yeah. So just try, I would say. Okay, yeah, good. Have a very good day for those in the US evening for those in Europe or Africa. And I would say good night for anyone in APAC calling in. Thanks for, thanks for being here and then we'll catch up later. Bye bye.

Thank you very much.