Webinar: Death of the Double Diamond and the new, AI-powered Stingray model

Hello, everyone. Welcome to today's webinar on the new kind of innovation Stingray model. We'll give it a couple of moments for folks to trickle in. We're currently reaching the 300 mark, so we'll give it a couple of moments before we begin. But as you're entering the webinar, can you just type into the chat what your name is and where you're calling from? And if you're feeling good, what, what you're most excited to hear about today?

Argentina. Brooklyn, Mexico.

Wow.

Budapest. Somebody's lucky. In Florida with some nicer weather. Sunny Edinburgh. Never heard that phrase before. Wow.

We have quite, quite an audience, Jeff.

Excellent. Wonderful. So welcome. Well, yes, welcome everyone. Thank you for joining us today for this webinar. We have an interesting topic, I'd say a bit of a provocative topic, maybe even given some of the chat we've had on LinkedIn and around some of the blog posts we do around this. And we're very much looking forward to it. Natasha and I will be your hopefully gracious hosts for this time together. We're very interested to share with you some of what we've been doing and have a discussion around it all. And I'll let Natasha introduce herself and then we'll go from there.

Hey, everyone. Great to see some of you again. And for the folks joining us or who didn't join us for the Autonomous Innovation Summit, welcome. I'm really excited to have you. As Jeff mentioned, I am Natasha. I'm an associate director at Board of Innovation, focusing kind of across industries, but really within the consumer space. And really with Jeff and a couple of other pioneers at Board of Innovation working to figure out what AI means for the industry. And so really excited to share some. Some thoughts that we've had around the innovation model. Jeff, would you like to introduce yourself?

Wonderful. Yeah, excellent. So good. See you all. Jeff Gibbons. I lead the Americas business for Board of Innovation. Quick background on us. We are a global innovation consultancy working with established companies to help them imagine new products and services and businesses and create them today. Sort of a bit of backstory on me, which I think is relevant for the story is I've been working in innovation consulting for about 16, 17 years now in the US but also a little bit in Europe. And I would say it's been a fascinating journey over the last year in terms of the explosive emergence of generative AI and how it's really challenged a lot of how we've looked at the role of innovation and the way that it works. This is something We've been experimenting and testing with a lot over the last year and we're excited to dive in and share more with you around what we've been seeing and I'm going to put on my screen share now, so I hope that is going to be working nicely. Today we have the somewhat provocatively titled webinar around the death quote unquote of the Double diamond model and talking a little bit about how we're seeing innovation changing today and the emergence of a new model that we have been finding to be more useful and more effective in today's world, which we have called the Stingray model. And we would love to have a discussion with you around what we, what we're seeing, what we're learning and the experiments that we've actually been doing in this space and open up the discussion around it. This is us, we've already said hello. What we're going to cover is a little bit of background on the context around AI powered innovation and really just assuming that most folks have maybe heard a lot about what AI powered innovation might be, but also some of the specific evidence and reasons that we have actually seen for the need to change the way that we look at innovation. We're going to talk a little bit about the Double diamond model and why we think it's losing ground and losing relevance in the current environment. Then Natasha is going to really go deep on this new model, the Stingray model that we've been experimenting with and seeing great results from, to talk about where we see its future potential and how we're using it today. And we would love to. We're going to keep it interactive, we're going to be doing some polls in the side chats and also please feel free to ask questions. Our colleague Christian is active in the chat and be answering questions and we'll have time at the end for Q and A as well. With that I will keep us moving forward. I think the first thing that's really important to emphasize is just that the emergence of generative AI really does actually mark a historic development in how we think about innovation. So I mentioned I've been working in innovation consulting for about 16, 17 years and when I started my career everything was all about design thinking and how do we be empathetic and human centric design methods. I would say in the 2010s everyone was then shifting focus to lean startup and how do we be hypothesis led and how do we experiment whilst being empathetic. But I would say the biggest change that I've seen in My career by far has been the emergence of AI powered innovation. And really, I think the potential that we're seeing is that it's not just about how do you tweak or improve the existing methods that people are doing, but actually how do you reimagine the role that innovation can play in companies and also dramatically improve its success rate. Because I think we have to be honest, as a community of innovators, that innovation in corporate environments has not been that successful in the last couple of decades and there's work to be done to actually make it more successful. So if we take a step back and think about what typically happens when a new technology emerges, like think about the emergence of newspapers online, basically what happens is that you kind of copy what was already existing in the previous context and put it on a new technology platform. So the first newspapers that emerged online in the 90s, it was basically just like a PDF, effectively of the newspaper. Same experience, but just in a new channel. How that's evolved, though, over time, as we've learned about the potential of web and mobile and how you can actually consume content in new ways, is that we've reimagined how you can actually deliver news. So it's much more personalized, it's much more interactive, there are different formats and ways to engage with news content, which many companies have been innovating with for the last 25, 30 years. And I think we've seen somewhat of the same thing with the emergence of AI powered innovation. So starting about a year ago, and I'd say this is from the perspective of innovation, we were there too. We started out with how we take the things we already did, like writing an interview guide, for example, and do it a little bit cheaper, a little bit faster, maybe a little bit better with AI. And it was, you know, quite astonishing to see how you could do that and how ChatGPT could be useful in all sorts of things. But what we started to look at now is also how can we actually take a step back and think about do we need to do some of the same tasks? Should we be actually thinking about new methods that are created and really fundamentally thinking about what is the role of innovation within a company and how can you actually reimagine how to run innovation in a company now that we actually have AI? And so that's more where we're looking to focus our efforts going forward. And some of what we're going to talk about today, and I would say we certainly haven't figured it out, but that is the mission that we're on as a company, and that's what we're excited to talk about today. If you think about what that means, there's a few things that come to mind, really. We'd imagine a situation where instead of people going out and having to empathize and spend time and weeks and months figuring out what needs are, you can actually anticipate and serve needs before consumers could even formulate them in their minds. Potentially, we could have a situation where concepts for new products or features could be generated and rapidly assessed across a range of different factors all at once. So in theory, it's possible to actually assess something in terms of whether or not it's desirable, feasible, viable, sustainable, and even what's the right market timing for a new innovation all at once. Rather than having to do this in a kind of sequential way involving lots of different people and ad hoc processes. Today, you could imagine a situation where AI is able to help prompt you and say, here's a new scientific breakthrough and here's how you could apply that to your business rather than you having to go out and do research on that yourselves. You can imagine a situation where products and services actually evolve along their lifestyle, along their lifecycle, to actually evolve and adapt to individual consumer needs. So imagine that a digital product like Netflix, for example, could actually evolve and adapt to my needs and how I consume Netflix content and Natasha's needs and how she consumes and uses Netflix content. Lastly, this whole concept of synthetic testing, a very interesting space of basically how you can actually use predictions around how people or systems would respond to individual questions, to situations, and how we can actually expand beyond that into entire simulations of business systems, of competitive environments. There's lots and lots of interesting possibilities around this, and we recognize that we're basically at the starting point of this journey. And I think as we've been posting a couple of things around this webinar and also having some discussions with other folks around this, we've recognized that as a company, we're kind of being a bit provocative deliberately around where we see things going and we know that there are skeptics out there, right? So there are. And even in the post we've been doing about this webinar, like there have been. There's been significant skepticism from folks around, is this really different? Is this stuff really going to happen? Can I actually do what you're saying and, you know, healthy skepticism? I would say that we're just being deliberately provocative. And I would say, clearly with the title of this webinar, we were being deliberately provocative. But it's also actually based on the experiences that we have. So want to be open around that. But the awkward truth actually is that we really see that a lot of this stuff that we're talking about is actually already happening. I want to share one example of one anecdote just to reinforce that actually as we're looking at the evolution of how we look at innovation, we have been looking at working with clients to try to put some of this model that we're talking about, this Stingray model, into practice. We'll get into more depth around what we mean by this. But effectively we have been working with clients to start to automate and partially automate parts of the innovation process. We've been working recently with a major client global company that probably most of you would have heard of to actually develop an AI powered copilot engine that actually runs the first two stages of this new model. We're going to talk about the Stingray model with human interventions at key points, but really actually with a system of prompting and synthetic testing and multi agent interaction to do a lot of the work that would typically involve humans. And I would say the speed of progress we've been able to make and the quality of the outputs has shocked us as a team and also our clients. Just to give you one example of this, we'll get more into the specifics of how this model works when Natasha gets to it. But we have basically in a two month period we're able to develop a piece of software that a user within the company would use and develop product concepts, assess them and figure out what would be effectively desirable concepts to start iterating and testing from. So like a starting point for those teams. And we took the outputs from this and we compared them against the outputs they had from actual real projects that they had spent a seven figure sum on with another innovation consultancy. And from the outputs of that, we actually took the concepts we developed with this Copilot engine with some human oversight, but very limited, and we took the concepts they'd actually done that they'd spent months on, they spent millions of dollars buying from another agency and we put them into a max diff test. And out of those 20 concepts, the AI powered copilot actually delivered four of the top 10 concepts and actually the highest scoring concept. And this is after two months of development and a very rudimentary system. And the point of this is not to say that what we're looking to do in this world is completely automate everything, but what we're saying is the proof is already there in the work that we're doing that it is possible to deliver, to deliver creativity, to deliver insight and deliver things that we know consumers like. And so we feel like it would be remiss of the world to not figure out how we can put this best into action. And so the point for us really is not to actually how we can do just that. We've proven that it's possible to automate what we've been doing for the last 20 years, to actually automate and figure out how to do the Double diamond model. But it's actually to figure out how can we take that and know that's possible, but use that to actually address the big problems in corporate innovation? So how can we actually improve the success rate of corporate innovation? How can we actually deliver more practical innovations, not just things that customers love? So, to be specific, if we've developed a website tool for a client that can develop concepts that we know consumers will love in minutes rather than months, that means those teams can actually focus on how do they make those things practical, how do they make them in their factory, how can they make money from them and actually focus on the more important work in their business, rather than spending weeks and months figuring out what a consumer will like? Really. The other piece I think that's really important is actually how can you help to eliminate some of the disconnects in terms of how corporate innovation works, to deliver closer integration between innovation, corporate strategy, and the institutional knowledge that already exists in big companies to actually make innovation more effective. So that really is our goal with how we're looking at rethinking the way that innovation works in companies today. So to get more specifically into the Double diamond model. So this is. We'll just do. We'll assume that a lot of folks have probably heard of it, but really this was a model that was developed back in 2005 and popularized by the Design Council and has been used by consultancies. And probably thousands and thousands of people have been trained on this model, and billions of dollars have been spent on projects using this approach. And we're not trying to criticize that at all, but we're just saying that right now we feel like it's losing its relevance in the current world. A little bit of background on this. The Double diamond basically benefits from its simplicity. It's a process basically where you figure out, first of all, what's the right problem to solve and narrow in on that first, and then you figure out within that problem what's the right solution or solution set that would actually make sense for your business. It's a simple way to conceptualize, to plan and run design innovation projects. A key part of it, as you'll see, the diamond aspect, is that you have this process of divergence, so identifying lots of different ideas for problems and solutions and then convergence. So actually narrowing, making decisions. And it's a very effective model. You know, for myself, it's the kind of thing I've been using for the last 16, 17 years in my work. And, you know, it's also, like, very. You know, I have many good memories and experiences of doing projects through this and, like, the deep empathy you have for consumers and customers and all that kind of stuff. So not to criticize it, but just to say that, like, we are seeing now some limits to its effectiveness based on the way that we see innovation happening today. And we had this autonomous innovation summit about six weeks ago, and one of our clients, Kuhn at PepsiCo, he said in five years, he thinks it might be that nobody is talking about the Double diamond anymore. And so that actually created a lot of chatter in the chat around this summit and really was kind of the impetus for doing this webinar, because we felt like it was worth a conversation specifically around it. So we want to talk a little bit about why we think somebody like Kuhn, who's an executive, a big company, is saying that and what we think could come next. And so there's a few things that I think stand out in terms of why the Double diamond is losing relevance. So I think the first piece is it's a very effective process for, like, structuring human thinking. But what we're seeing is actually that success and innovation is no longer just about human thinking. So complex problem solving is now really about for companies managing the complex interplay between human thought and artificial intelligence and data that exists in the company that's often unused and unlocked. And so that has some practical implications. So instead of spending weeks weighing a handful of problems or a handful of solutions at a surface level, that because we as humans can only think about, say, seven or 10 or 12 things at a time, we can actually do things differently. We can consider dozens of ideas or hundreds of ideas at once. We can actually think about problems and solutions together. We don't have to think just about a problem and then just about a solution. It's actually much easier for AI systems to actually think about more complex things in ways that humans can't actually do them. The second piece is that unfortunately, the practicality of the way that the Double diamond works, regardless of these specific methods you use, is that often what happens is that people end up focusing most of their resources and their time on this question of desirability. What does the consumer or customer want and how can we best meet their needs? And so what that means is that teams can often spend weeks or even months interviewing customers, doing empathy research, running workshops to kind of synthesize all this stuff. And that's great, but too often, unfortunately, and I think as somebody who's worked in innovation consulting for a long time, I've probably been part of the problem myself. You end up spending weeks and months to deliver ideas for new innovations that maybe can't be made in that factory. You haven't actually spent the time or had the time to think through how can we make money from this? And this is actually a real problem for the innovation community, I would say. And what we found is that with generative AI tools, we can actually get much faster to an idea and a concept that is desirable and spend much more time focusing on the more important questions I would say around how can we actually make things feasible, how can we make them viable, how can we actually cut to the chase effectively and skip some of the unnecessary steps to focus the resources on actually making innovation successful? The last piece is an interesting one and maybe just a little bit unexpected. I think when people talk about AI, they often talk about biases in AI systems. And it's absolutely true that AI systems do have their own biases. But I think it's. And just to give you an example, you think about text image visualization models, there's a massive amount of bias in terms of how they actually visualize and are built on stereotypes effectively in the world around people, around ethnicity, around gender roles, all these types of things. I think the one thing that's really important to remember in all of this though, is that the reason that those models are biased and sort of these early models, I would say, is because of the humans who created the data that they're trained on. Right? And so actually we see a much bigger opportunity to actually think about how can we actually debias the way we work using AI as a counterpoint to the biases that we as people actually bring into our work. So if you think about how Double diamond model works is very built around human decision making. And there's a lot of things that happens when humans work together, right? So we kind of struggle to make sense of lots of information all at once. So we might cut corners, we might fall prey to our own biases. We might fall in love with solutions too much because we just have a personal attachment to them. We make shortcuts. We only focus on kind of like a mainstream user because we have to fulfill a certain quota of 20 people that we have to interview. So we don't actually think about doing interviews or research with underrepresented users. This is kind of too complex to do all of it all at once and you have to make resource trade offs. And we know as teams like people get anchored to their own experiences and stuck in established thinking patterns. And so really what we're trying to look at with the way we're using AI now is actually as a copilot for teams to actually overcome those biases and understand how we can actually retrain the way that we work. And I think it's really important to emphasize that it's way, way easier to actually retrain an AI model that has been in existence for six months or a year, or it could be a brand new model than it is to actually retrain a human being like myself who has life experiences and existing patterns of thinking that are going to take serious effort to actually overcome. And so we see a lot of potential for actually using AI to overcome human biases and the shortcomings of the human mind too. So with that, we wanted to just have a bit of a discussion around these three points we made that success and innovation is not about human thinking alone, focusing resources on desirability rather than feasibility and viability. We're going to do a poll in the, in the chat here and see which of these resonates with you most as a group and then Natasha is going to take us forward. Okay, so I see some, some votes starting to trickle in. So it seems like we've got a. The leader seems to be that success innovation is not about human thinking alone. Less, less love, I would say for the, reinforced the reinforcement of human biases and shortcomings. And yeah, I see an interesting comment from, from the chat around the innovation sweet spot is where desirability, viability and feasibility intersect, which is actually a perfect segue. So I think where we're trying to move with the Stingray model to actually try and address all of those things at once. Excellent. With that in mind, we will shift gears now. Natasha is going to take us forward in terms of how we've been looking at this new Stingray model that we've been experimenting with and how it's been working out.

Awesome. Thank you, Jeff. As Jeff has mentioned, there's kind of this new emergence of a new model that just helps us rethink how we kind of can do innovation, get rid of some of those efficiencies that Jeff was speaking about earlier on. And it's really thanks to the emergence of AI and we really see kind of these efficiencies in the market that Jeff was talking about kind of directly in the market. So many companies have redirected their resources from innovation to other parts of the organization. They've downsized teams, they've closed innovation labs. And so in turn, it's really created this added pressure for innovation teams to be more effective, to be more efficient, to improve their success rate to the market. And so with that in mind, we developed this kind of new stingray model. We've developed it, we've tested it, like Jeff mentioned, with companies like PepsiCo and other companies out there to really see how we can kind of overcome these hurdles at the double diamond or kind of legacy innovation methods have created and try make sure that we can ensure that we're more kind of efficient essentially. And so with that, there's a couple of benefits that we'd like to run through before we dive into kind of the meat and bones of the this webinar. But as kind of Jeff mentioned, there's a rethinking of resource allocation. So think about kind of the time, the money and the effort that you put in to your current innovation processes. We're really saying that we can redistribute kind of your time, money and effort when you're increasing your confidence in a potential solution. And that's what the stingray approach helps us do. It's what kind of an AI and human and AI approach effectively allows us to do as well. The next benefit that we wanted to speak about was that we usually focus really on the empathy of customers and consumers. And that is really truly a very important part of innovation and will remain regardless of the onset of AI. But we're really seeing kind of a shift focusing on experimentation. And we see that because empathy is a little bit more passive and it's a little bit more actionable. And if you're not doing something with that empathy that you create of a moot point. And so the Stingray really focuses on experimentation to get real life proof points of successes and failures of the concepts that we're trying to test. And through that kind of empathizing with our customers in a more real life manner, we also see that we're getting to solutions a lot faster. And so not Only does kind of more experimentation earlier on in the process lead to solutioning faster. But the Stingray approach also brings in the lens of viability and feasibility. I think we had someone in the chat say that innovation happens at the center of those three. Viability, feasibility, and desirability. And that's absolutely true. And so we're really bringing kind of the things that we think about at the end or in the back end of innovation, really in the front end of innovation. And that helps us get to more validated solutions and more effective solutions a lot further faster. And as Jeff mentioned, I won't harp out about this point, but it really kind of helps us overcome human biases. Opening up kind of the aperture to consider multiple problems, leveraging our human experiences, but then also thinking beyond that, essentially, and kind of everything together allows us to kind of overcome and tackle more challenging problems, creating more deliberate goals, essentially, as we are innovating. And so with that present to you kind of the Stingray model, it does seem that a lot of people have seen this model before or the visual of this before. And I just want to emphasize that, yes, it does look like a resizing or a side iteration of the two diamonds or the double diamond that we see today. And there is definitely this element of diverging and converging that's still super important within innovation processes. But that's not necessarily to the point. The point really is, like, what do you do at each of these different phases, how do you progress through the model, and really, where do you land up at the end of the Stingray that differentiates it from the traditional double diamond approach. And so we see that there are three different parts to the Stingray approach. There is a train, the develop, and then, of course, the iterate. And we'll dive into what each of these sections really mean. And then we'll also go through an example of building toys for. For children to really bring this home. Great. So diving into the train, so similar to kind of innovation projects, your team will start off with a set of goals for the project. So you'll define kind of what your parameters are, constraints are, what types of solutions, what types of problems you're interested in solving for, and what kind of our metrics of success. That's pretty kind of stock standard that we see today. But then it really goes into gathering relevant information. So usually with the double diamond, you'd want to understand and empathize with consumers almost immediately here we're saying, let's take a slightly different approach. Let's gather kind of Relevant information that will help us define our problem and hypothesize our problem. And that does include kind of consumer data trends with our consumers market trends. It's also leveraging kind of your company proprietary information that you have. It's also bringing in kind of feasibility lenses. So what is technically feasible in our, in our factory, for example, or what do we have to consider from a viability perspective? So it's bringing all of those various different perspectives together. We're also really lucky with generative AI that we have a lot more information at our fingertips. And so training this model becomes a lot more quote, unquote, easier for us to do. And so we'll take this kind of copious amounts of information and we'll essentially train that model. And that means a variety of different things. Right at the lowest feasibility level that all of us can essentially do today, we can take a lot of information and train like various chatbots that we can find from OpenAI Anthropic or Google or kind of the other likes of chatbots. Right, that's like your lowest level of training and model essentially. But kind of the more high fidelity models that we've been creating, and Jeff also alluded to earlier on today is really creating kind of company specific LLM models or various kind of engines, for example, that can essentially be replicable beyond the project that we're working on, for example. So you basically take this information, you train a model, and through training this model you're able to essentially generate and prioritize kind of problem spaces. And so at the end of this entire phase, you really are getting, you really are developing kind of guidance for your team to kind of allow them to understand different perspectives and tackle the right problems. And so you're developing and you're delivering kind of solid kind of prioritized hypothesis on problems and solutions. And just a quick note that this model doesn't necessarily have to take weeks or months. You can do this within days or hours, just depending on kind of the human involvement that you want to include in this model. And so with that, let's kind of move on to the second step, which is develop. So within this kind of develop stage, we're really kind of ideating very similar to kind of the double diamond. We want to get kind of want to diverge and get kind of a range of problems, but instead of kind of ideating maybe like 30 or 50 ideas just through human intellect, we're really kind of combining AI and human either individually and separately or together to develop kind of dozens or Hundreds of ideas. The other thing that we could do is kind of go beyond what we can do as human thinking, and we can actually ideate on specific problems that we have. We can ideate on specific trends, think about kind of in the future, we might get to a world that does a specific amount of things. We can actually ideate directly on those trends and that future world that we're in. We can also take kind of feasibility into consideration and say, with these kind of feasibility constraints and with these consumer problems, what can we ideate on? And so it just really opens up the ability for us to ideate beyond what our kind of human capacity allows us to do. And so really at this stage, we're kind of diverging much larger than we would in previous kind of human only methodologies. Another kind of point to make on this stage is that we don't necessarily need to consider problem first and solution, as Jeff mentioned before, which also means that we don't necessarily need to think about like specific opportunity areas, because opportunity areas is traditionally a means to an end. It helps us to kind of direct where should we be ideating, what solutions should we be, should we be developing. But really, with the onset of AI, we're able to kind of consider both at the same time. So we're making the way that we do things a lot more efficient and skipping certain steps just because of our increased capacity. And like I mentioned before, the true magic really occurs not only when AI kind of generates hundreds of ideas for us, but when you combine it with human intellect, stretching AI ideas with kind of human thoughts, with our human lived experiences, or doing kind of a combined AI kind of and human and human thought ideation. So that kind of then takes us to the iteration phase. And this is kind of a longer phase as kind of the design shows. Another thing about kind of the design just to kind of make sense of it is the darker we get, so the longer we're kind of iterating on something, the more confidence we get with a particular solution. So kind of bringing that and embedding that into what this phase actually means is we've landed on the previous develop phase with a bunch of various potential solutions that we have developed. And instead of kind of prioritizing it down to a select few that we were able to kind of take forward and kind of feasibly learn more about, we're able to kind of consider all these ide all at once. We're also able to kind of bring those ideas to life. So we're able to ask AI to kind of develop concept statements, to develop specific features for those concepts, to develop images for those concepts. And that then essentially helps us to test these concepts and these ideas that we have multiple times over so that we increase in confidence as we go through this iteration phase and what we're suggesting. And of course, every single project is different in its own right, but really we can think about at the early stages, when we have so many different solutions or solution ideas, at least we're able to kind of synthetically test those, which is a cheaper, more efficient way to do so, especially when they're initial ideas, when we're not necessarily that confident with them. So we can train our models in that first stage to take the form and the experience of our consumers, for example, and really test it synthetically with them. We can also bring feasibility and viability into kind of the synthetic testing. So which ideas actually aren't feasible within the horizon that we've chosen to kind of test in? Or if it's not feasible, how can we iterate on that solution to make it feasible and make it kind of relevant for us to pursue further? And as we kind of go down this funnel of iteration, we're able to kind of slowly but surely incorporate human interventions, essentially. So, for example, you can progress to a AI and human experimentation approach. So for example, that could be an AI chatbot conducting multiple empathy interviews, which means that you can essentially interview hundreds of people over a weekend, over a couple of days, versus kind of human beings interviewing, let's say, eight to 10 people across a week or five days. And then slowly but surely, as you're gaining more and more confidence, you're able to do more human intervention. And so you can prototype the product and just do kind of consumer testing, essentially kind of person to person testing kind of two points on this. The one is with this reallocation of, of resources and the thought of shifting our, shifting kind of how we test depending on our confidence level. So the more confident we are, the more human testing we can do. What that thinking actually does is it allows us to also actually go out and speak to people who we might not have considered to speak to in the first place. So during that synthetic testing phase, for example, if we want to consider people that we've never spoken to before, maybe those at the extremes or the fringes, we can essentially test the genpact with synthetic synthetically. And then we can as human beings kind of go out and test what might be happening with populations that we might not have considered before. So it's really Kind of this iterative round of testing that you need to develop, depending on your. Depending on your. On your project that you're working on.

Great.

So with that, I'm going to dive into an example. So the example that we have really brings to life this stingray approach, and it's focused on building toys for children who need to kind of develop their cognitive functions, for example, and support them with various kind of learning disabilities. And so we start off at the train phase and we kind of set our objectives. And in this case, it's kind of creating a toy that is not only educational, but really appeals to all children, so it's accessible to all children. And we've kind of just defined for illustrative purposes, a couple of success metrics here. We then get into the information gathering phase where we're collecting a bunch of data. So from this perspective, we want to gather data from, from parents, from children themselves, from experts, from our competitors, for example, from various kind of market trends. And the list can really go on and on for this phase. And then, of course, we kind of train the model. And once you've kind of trained that model on the kind of information that you have gathered, our output really is a prioritized set of problems and design elements that we. That we want to kind of take forward. We then get into the develop stage. And so with the model that we have, we're able to kind of ideate dozens or hundreds of solutions. And so we have a couple examples that AI generated for us very, very kind of briefly. So we get like modular puzzles, for example, we get kind of augmented reality, various kind of solutions. But we're really kind of being super expansive to develop these ideas, leveraging not only AI only kind of ideation techniques, but then also kind of human ideation. And then, of course, we can use AI to prioritize some of these solutions. So, for example, we can divide them up and group them into kind of mutually exclusive groups, or however we decide to do that, prioritization. But we can basically get AI to organize these hundreds of ideas that we have, synthesize them into ideas that kind of overlap, and really help us organize so that we can continue our thinking as we continue along the process. And so the output of this section really is a broad and kind of categorized set of solutions. We can also bring solutions to life at this stage. So we've got kind of concept statements and like visual kind of, of 2D prototypes that we can generate using like midjourney or dalio, any of these other text to image tools. And then last but not least, we come to the iteration phase. And so starting off with kind of synthetic testing, moving across to human and synthetic, and lastly, of course, to just human, we can see that we can synthetically test with, for example, parents and experts. That's who we choose to kind of synthetically test with, iterate the ideas as we learn from the feedback that we get from these synthetic Personas or panels that we've essentially created. And then we can go on to kind of human and AI testing. The tool that we use quite often is a tool, like I mentioned before, it's essentially an AI chatbot that we engineer our kind of qualitative questions to. And this AI chatbot is able to talk to real human beings and essentially gather kind of qualitative data from those, those people, either in text form or in kind of voice form as well. And then last but not least, we can essentially prototype a solution. And the example that we have going on on this slide is an example of kind of building blocks with braille and audio descriptions as children are playing with it. So with that, that kind of takes us till the end of the Stingray approach. And I'll hand over to Jeff, who will take us home.

Excellent. Thank you, Natasha. And recognize we've had a couple of complications with Natasha's wi fi. Hopefully mine is a little better. So I wanted to just round out with a few things. One is just sort of a little bit of a sense of how we're starting to use this Stingray model. And to be clear, we're still experimenting with it and then also address some of the questions that I've been seeing in the chat too. So the way that we're using this today is we're using this actually to fuel the current projects that we're doing with our clients. So actually working with them to figure out how can they actually change their ways of working, how can they accelerate the success of their projects. To do this, we're actually using, working with clients, also build some custom engines for them that actually do this work, which actually gets us to some of the questions around how is this actually working in practice and what's really possible. I'll maybe just touch on a few questions that I've seen coming up. A lot of questions around what tools are people using for this. I think what we're seeing is a general shift away from just using sort of off the shelf, publicly available stuff like ChatGPT to actually teams and companies actually building their own custom engines built on multiple models and or new models that are proprietary models rather than just working with what exists from OpenAI. So actually building models that are trained on their own proprietary data as well, whether that's customer data or things like the capabilities of their factories. There was a question around what do we mean by training? Corrected is a very broad term for us. What that really means is it's a mix of prompting, pre prompting, fine tuning and rag methods typically. And actually we've been working with clients to actually develop the sort of infrastructure and models that they need and the interfaces to do all that. And as I mentioned, that's usually using existing models, but increasingly seeing a shift towards specific proprietary models that they actually develop themselves for various reasons in terms of avoiding ownership issues around what emerges from them as well as even just from like a sustainability standpoint too. Couple of questions around, are we eliminating designers? Is everyone losing their jobs? I personally don't think so. If you look at most technologies actually what happens is that you just create lots of new jobs. So like if you look at the example of the spreadsheet, everyone was worried that accountants would lose their jobs when the spreadsheet came out. What's happened is that now there are way more people who use spreadsheets and do much more complex work in them. I think the same thing will happen with AI and we're going to see that innovation teams are going to be different in terms of the types of capabilities. But I don't think designers are going away. There was a deliberately provocative point in there around experimentation over empathy, which clearly ruffled a few feathers there. Really the point here for that one was really that the kind of passive side of empathy and just sort of understanding people, I think we're seeing less relevance moving forward from that. Frankly, just because I don't think most companies are interested in investing in the, in the sort of personal experiences of their innovation teams. I think they're most focused now on how can they actually get the results from innovation. To be totally transparent from what we see with our clients and good question around the real world example. And as I mentioned, this is like fairly early days. We're sharing a model which we are experimenting with. We would love actually very shortly to share a more practical example of how it's actually working in practice. And we will do so as soon as we actually have something we can share without violating client confidentiality. And maybe just the last point wanted to share was just there was a question around what does this mean for qualitative research, quantitative research, the role of those things? I don't think any of those things are going away? I think really the focus is more actually how can you relocate your time resources to get to focus those in new ways? So for example, if you're able to figure out much quickly a starting point that you know is very useful and desirable to a customer, then you can actually, you can actually really figure out what's the best way to actually use qualitative research to figure out if this is sustainable. Can you actually make this, Will this actually work in your industry? Will customers actually adopt it? All those types of things I think would be where we would see more focus shifting over time. Couple of last things to close it out. We will be sharing the slides from all of this. We look forward to keeping in touch with folks around this. If you could shift us to the next page, Natasha. Yeah, so we will be following up with an email which gives you the slides and you can actually, I think you can download the slides from the documents as well. And then you can also, you'll get a recording of the webinar as well. We have another webinar coming up on the 1st of February around how we can actually apply AI to next generation approaches for venture building. So like a more specific example of how we can look at that in new ways. And obviously we are a consultancy and we work with clients all the time on these types of things. And if you're interested in having a conversation around how you could actually demonstrate this type of model in your company, please get in touch. And I want to thank everyone for the time today, enjoyed the chat and look forward to continuing the conversation.

Thank you so much.