 This is a webinar, more like a conversation, I would say, around the topics of how can we incorporate AI and predictive analytics in decisions that have to do with suppliers, with food suppliers. And it's a very special opportunity to have this conversation, because when we dive into the details of specific decisions or specific tasks that people are doing in their job, then looking at the ways in which digital technologies can help us, and reveal lots of information about things that we have thought of perfectly well, or things that we haven't thought at all. So I hope that with this wonderful panel, we'll be exploring what can be done and what cannot be done, but should be done in everything it has to do with suppliers and their food safety performance. I will officially pick off. I've already introduced the topic that we will be covering. And it's an important one, because especially during the past couple of years, we have been listening to several people in the food industry, and in particular in food safety and quality discussing about the promise that predictive analytics and AI is bringing. We have also heard lots of skepticism and criticism about whether this is a technology that really adds value. People have concerns on whether this is a marketing bubble, period. So these are some of the reasons why we took the decision to organize this session. And I thought of having this session by getting some people, some experts around the table that can bring the different perspectives that we can be. And we should be listening to it. So each one of the experts that we have today, they have been carefully selected for a reason. I will start by introducing Takia, Takia Vol, with a PhD in food safety microbiology. Takia, you have spent lots of years also working in surveillance, microbial surveillance. You have been responsible for laboratories, USDA laboratory, looking at risks like Salmonella and Picolai. Now you work for Sargento. Hello. Yes, yes. I've been with Sargento for about four and a half years now as a food safety microbiologist and looking forward to this discussion. Wonderful. We also have Dennis, Dennis Tracy with us. Dennis is a leading cultural compass and organization that he founded and has set up to support other organizations or teams when they want to transform and move into different cultures in terms of how we respond to safety or crisis issues and the way that we manage these issues. And there is lots of experience that you're bringing in this, Dennis, because you have been for many, many years in both scientific and executive and leadership roles in the food industry. Welcome, Dennis. Thank you, Nikos. Yes, when I think back to my first job in 1979, working as a microbiologist in the laboratory, the journey's been long and it's been eventful. But I've also been a manufacturer manager. So I've run factories as well as been in global supply chains. And my last role was global president of quality safety security environment for Plattice Global, which is a $5 billion snack food business based across 36 countries. And our products touched 4 billion consumers. So half the world's population was enjoying some of our products at some time. So my responsibilities were quite focused in the area of food safety and health and safety, of course, as well. So thank you very much. Cool. And we also have Michalis. Michalis is leading our data services team and he's very special because he's one of the people that has been working with data that is out there, everywhere out there, that we can identify, scan, collect, and combine so that we can then inform and train AI models so that they can generate predictions. And he has been doing that as well. He has been one of the first people that I have seen doing actual hands-on work in getting the right data together, setting up an AI model and then trying to generate predictions that make some sense for food safety professionals. Michalis, welcome. Thank you very much. Thank you. Very nice being here. I indeed, as you mentioned, my background, first of all, is in computer science. Throughout, I think my career, I've dealt with software-related and data-related problems. And over the past eight years, I've been with Sagrano working both as a data engineer and also leading our data team. Great being here, Michalis. Michalis, why are you here today? What is your expectation from this conversation? I think my expectation is to have a very nice discussion around data. How can data be used to train AI models and how these AI models and the child foods can be used to make our data decision around food safety better? And Dahlia, what about you? Me, I'm mute. Yes, same as Michalis, just to hear about the data that is out there and how we can use to enhance our monitoring of what external issues that are going on around not only the US, but globally. Dennis? Nikos, my experience has shown me that data is invaluable. Data that we know, information that we know is invaluable in helping me to make decisions. The idea and the concept that AI could be used to enhance that knowledge is something, even for an old dog like me, this is very, very exciting. So I'm here today to find out as much as I can about what AI can do and how that can help organizations stay ahead of challenges. And I hope that I will play successfully during our webinar. I will be pushing you to keep in mind that AI is not only about what has happened and we know, even if it is something that has happened a few seconds ago when it's magic that we can find out right now. But we have to keep in mind that AI is a lot about things that we can estimate or forecast or expect for the future. And how does this change the way that we do business? How do we take decisions? And together today, it's all about decisions that are related to people, organizations that we work with, our suppliers. So it's even more sensitive. Let me take a quick pause here as we are talking about AI to ask our audience, what do you know about AI and how much do you trust? Understand and trust AI. There is a team hidden behind the curtains that will take care about setting up a poll. Let's see how we will do this. This is a statement. I fully understand and trust AI technologies. What does our audience think? How do you feel about these technologies? Do you also see the results as they are coming in? I don't see many people saying that they strongly, fully agree with understanding what this innovation is about. Many, many people are neutral. And lots of people really don't understand or don't trust this innovation. OK, that's interesting. That's very interesting. Let's keep this in mind. And it's interesting and relevant to all conversations that have to do with change management and introducing an innovation in any process or way of doing things. There are the people that will jump into adopting any innovation even if it's not working. And then there are the people that are more skeptical and they want to see the actual value that they can get before they invest their time and their resources in innovation like this. OK. So you see the results, I hope. I have a small intro that I would like to give to put a bit of context. I'm doing this quite often and it's going back to the basics. When we talk about artificial intelligence, AI, what does it mean? Is it some kind of black magic? Someone is doing something with a crystal ball and string things appear and then wisdom is coming towards us. Well, by definition, no. AI is a branch of computer science. So it's science. And there are some processes and some technologies that we put in place so that we can make a machine, a computer, able to imitate what we consider intelligence and behaviors of humans. And in many things, many of these behaviors or intellectual processes that we do as humans, we put AI in practice because it can do things either a little bit faster than us or because it can process much more information in a much more accurate way compared to us. Or because a computer can store and continue to grow and develop upon experience that has been built throughout the years, which is very often lost when you have someone, a human expert that leaves a professional post and then someone else starts pressing. So what we see when we talk about AI and systems that are trying to imitate intellectual processes, we talk about things discovering, like discovering meaning into lots of data or generalizing from lots and lots of observations or building upon past experience to extract something about the future action. I really like the weather forecasting analogy because I remember it as a kid. Every night, we were waiting for the news to look for the weather expert that would tell us in a few minutes what we should be expecting for tomorrow. And lots of our decisions were affected by these expertise. And I would say that even today that we do see those experts trying to communicate and share with us what they see in these powerful models, forecasting models. Sometimes we still have them. We still rely upon them. And sometimes they're not there. The forecasts are in our fingertips, at our fingertips, in our mobile phone, in our computers. And we have to take the decision based on what we see in this system. This is the type of conversation that I was hoping and I'm hoping that we're going to have today because there are lots of decisions associated with preventing, understanding, mitigating, quickly reacting to food safety risks. And there are different steps in which we can go and look at and see where we can apply some of these predictive technologies. And if we understand very well the types of decisions that we have to take at every given scenario, then we can think about the best way in which we can put this technology to work for us. And this ranges from very, very traditional but very time consuming tasks like risk monitoring, looking out there about everything that is being announced and understanding if something is relevant to my supply chain, my organization, the so-called horizon scanning part. It also has to do with how do I combine data that is coming from different sources so that I can do risk assessment of ingredients, of suppliers, of a process, and go to other types of decisions like, where should I be allocating my budget? Where should I be sending my internal auditors? Where should I be hiring external auditors in order so that they take place and so it goes? That's my first question for you, my dear panelists. When we look at supplier-specific decisions, which are the types of questions that you see that are more important than relevant? Which are the decisions that are very, very important based on your experience and your day-to-day practice? Like you have left here first. Sure. So what we want to know about our suppliers, first we look at our selection process. As part of our supplier selection process, we're looking for a full risk profile of the supplier. And this includes the supplier itself, the specific manufacturing location, and the material they are producing for us. And then we would like to see the historical incidents of the ingredients and being able to take this information and predict the likelihood of incidents happening in the future will allow us for better monitoring of our suppliers and help us prioritize the frequency of our audits for our suppliers. And knowing the type of risks that implicate the ingredients will allow us to strengthen our audit programs and process controls and our environmental controls. For our high-risk suppliers, this can help us identify mitigation strategies to control or reduce or eliminate the risks. This could be either beneficial for them or us or not. If we know there is an increasing risk, we can work with the suppliers to put plans in place to mitigate those risks. Most of our suppliers are willing to improve their controls to avoid having any impacts on their products. And so the forecasts of the risk would affect them more in a positive way. And that's what we would like to know from our suppliers. That's very important. So what I hear you say is that this is a collaboratively process, high-risk identification. And the decisions always affect and involve your suppliers, right? Yes. That's very interesting. Then what do you think? OK, so from my perspective, my objective is to maintain the revenue in my business and to remain competitive. So what I'm looking for is repeatable and predictable outcomes in anything that I do. So for that, when I look at sourcing my materials, obviously I'm looking at, of course, we'll risk assess all of our suppliers. We'll risk assess the geographies. But when I send auditors to my suppliers, they're looking for three things. They're looking for consistency in food safety, of course. They're looking for the ability for that supplier to continually deliver that material. Can they make it consistently to the right specification? And thirdly, we're looking at ethical operating. So do they represent my brands and the brands that I'm trying to, the brand values that we project? So in all of that, the assessments are based around both the geographies, but also the material groups as well. So I'm looking to understand that my supplier knows their material as well. So do they understand the risks associated with their particular material? Do they understand how to mitigate that? And I'm also looking to build relationships. So do they understand that if they are operating within my business that they need to communicate to me when there are problems? So when there are issues in the global sourcing environment, I expect those people to communicate that to me before they communicate to any other supplier. And those things take knowledge of the supplier, knowledge of the supply base, but also the understanding of their insights. So can they share knowledge that they have about that material group or geographies or issues that may occur? So anything that adds to that insight, anything that adds to that knowledge, it gives me a competitive advantage to remove issues or to predict issues. And as Thakanya says, mitigate issues is intensely valuable to me. So what I hear you saying and highlighting a lot is the importance of having a process in place that will ensure again and again and again by well-defined steps that everything is within specifications. And risk assessment in this context is a repeatable step in such a process. It's not one of alarm that goes on and we start firefighting. This will happen. But this is not what you're looking for. Ideally, you're looking for something that can be incorporated in the process to make sure that everything is done properly. Do I get it right? You do. But that includes the suppliers that I have a relationship with but also suppliers that provide material through agents. So my business may source on the open market. So I want to have suppliers that understand that market as well because there may be issues or situations globally, pandemics, geopolitical shift changing. A country can change its sourcing through an election change. So all of these elements we need to understand and we need to manage for. That's very interesting. That's very interesting. Let's see what our audience thinks about the importance of questions or decisions. How do they prioritize these type of questions that we are putting on the table? Again, we have selected a couple of the questions that could be answered with the help of AI. And we are asking through Paul the participants of the webinar to which are the questions that you believe are more relevant and more important for you. You can choose more than one options. It can be very focused, like allocating, prioritizing a batch that is coming from a risky area and sending it for additional testing or including a new type of test for a particular contaminant when we know that something is happening in the market. There could be some decisions that are very, very radical, like we don't want to work with a particular supplier anymore. Do you guys see the results as they come in or do I only see them and I will share them at the end? You don't, eh? So I see which are the questions that are very, very popular. That's interesting. And I will share the results with you now. Everyone has answered. Here you are. Lots of people are considering the emerging new types of hazards or contaminants that should be informing their testing plans. Geographies matter. Should we be looking at different geographies because we see increased risk in particular geographies that we source from? Keep this in mind because we have something interesting to share at the end of the webinar. The last question, if we should be revisiting our decision to work with a particular supplier, seems to be also very popular in our audience today. OK, let us now go to some actual use cases, something that we can do today with the technologies that we have at hand. And this is where I'm asking Michalis for his help. Michalis, I would like you to walk us through these three use cases. Let me know when you want to move from one use case to another. First, thank you. Of course, Nico. Thank you very much. So what we have prepared for this next part of this webinar is going through three specific use cases where we see the application of AI in supplier risk assessment coming into play. Before we start, however, going through the actual use cases, keep in mind, which as is the case for pretty much every AI model out there, the important thing, the important first step before training any kind of model is the data availability. What you are about to see, first of all, are actual screenshots from our software reserve solution, Fudakai. And second, more than a million different data records have been taken into account when performing this kind of calculation. So let's go through the actual use cases. Let's start with the first one, thank you very much, Nico. And it's how can AI assist in performing supplier risk assessment for specific ingredients? What we are seeing here is, again, actual screenshots taken from our system. And we're starting here the live risk profile of a specific supplier. You are seeing on the top side of your screen a risk assessment performed for this specific supplier on three different aspects based on the ingredient. One is sourcing from them based on the number of incidents, food records, and body directions they've been involved in, and finally based on the count of operation. This is on the top left side of your screen. On the right side of your screen, you're seeing the actual data behind these calculations that are taken into account when performing this risk assessment. And now the interesting part here is this highlighted block we have on the lower side. What we have done here is out of all of the ingredients one may be sourcing from a specific supplier, we are training our forecasting models in order to identify which are the hazards that are expected to be on their eyes over the next 12 months. And I believe, as a matter of fact, that this specific use case may fall well in line with the results of the previous poll that the audience has answered. Because one of the answers posted by the audience was around being informed about imaging contaminants coming from a specific supplier for specific ingredients. This is something that could help identify such cases and focus or prioritize the audience and lab tests performed mostly on these specific ingredients as opposed to others that are expected to have a lower risk. This is as far as the first case goes, Miko. What you're saying, Michalis, if I get it right, is that this view is not only providing us with a forecast on whether a supplier is of higher risk or expected higher risk, but is also giving us an opportunity to understand which are the emerging increasing contaminants that we should be expecting as far as this particular supplier is concerned. Okay, please, dear panelists, keep this in mind. I will ask you for a reflection after Michalis is done and we're moving to use case number two. Perfect. So for the second use case for today, what we have focused on is performing AI techniques and training machine learning models to focus on a risk assessment coming for the previous coming from specific geographies and or for suppliers based in specific geographies. Again, we're taking into account all of the historical data we have at our disposals and we're training these forecasting models in order to identify if we focus on the image on the top left side of your screen, which are the countries or regions that are expected to have an increase in terms of risk assessment for ingredient eggs being sourced from there or for suppliers with companies based in there. So we can use this kind of data to train our models and identify which are the riskiest let's say agents for this period of time and over the next 12 months, as opposed to others that are showing a lower risk. This is as far as the image on the top left side is concerned and what we can do is also go into further detail, dive in deeper into the specific geographies that are of interest to us, whether we are sourcing ingredients from there or whether our suppliers are based in there. We can dive in deeper into the expected, the forecasted risk assessment profile for a specific geography. In the lower right side, where the showcasing an example here, what is the expected number of incidents and in respect to the risk assessment for nuts and product seeds originating being cultivated, produced in Argentina. And you can see both the expected values on the right side of the image, as well as what our most accurate models believe I've talked in the past. So Michael is here, what I hear you saying is that this is a view, a use case scenario that is looking at geographies from where we are sourcing and especially I can think of a scenario in which we are somebody working with foreign suppliers in countries where they feel less secure about or they're a little bit more far away from and it is important to have an understanding or an expectation of what might happen in terms of this particular geography. Keep this in mind my dear panelists, Sarah, because I would like your reflection when Michael is done. And number three. Perfect, thank you Nicos. And now for our third use case, what we've seen so far is how can we use the historical data disposal to identify out of all of these applies we're working with, which is the one that's expected to have an increase in terms of risk over the next three months or which are the regions one should be on the lookout for again over the next year. However, as I'm sure people in the audience are very much familiar with, most food companies already have in place some internal supplier management systems or their own internal power BI or in general BI dashboards that they use to make the data decision process easier. Third use case we have identified here is that these data, these insights that you've seen in the previous two use cases, along with all the rest that can be generated by training AI models can be used and can be incorporated within within these internal supplier management systems where the rest of your own, of your internal data around supplier resides. In this case in this screen, so that we have here, it's coming from third party solution around supplier management. What our own have done is that you have the overview, the picture, the data, your internal data in the center, but you can also include data coming from different systems. In this case, from our own data, as a service solution risk data, but of course the data provider can be any kind and such kind of system that can support these operations. So you can use these insights, incorporate them in your own supplier management systems and be early alerted using the tools, the software platforms that you are already using in your everyday life. Because I hear you saying two things that are important here. First of all, that in this scenario, what you're describing is how forecasts or predictions can be integrated in another platform, not the platform that we are developing, but another platform. And then there is an API that is serving this to the other platform. There are ways in which they can be integrated. And you briefly touched upon the importance of having this presented along with internal data that the company wouldn't share with the outside world. And this can open other interesting directions of work in such an integration. Okay, very specific, I hope, very clear scenarios presented by Harleys. The time has come for our panelists to tell us what do they think? So what do you think? Which ones or which one did you find practical and relevant? If something is crap or too far-fetched and what else you would like to see that is not presented here, right? What do you like to start? Yeah, sure. So I'll probably change a little bit what's on the screen, but yeah, so the first case is very relevant for us. As most companies, we do our risk assessments every year. And to have that information to help us with those risk assessments is very, it just will help us know where the challenges are with particular ingredients and what we need to do as far as do we need to increase audits for those particular suppliers with those ingredients or increase process controls on our end to see if we need to do more when there is a higher risk for a certain ingredients. I would not say anything was far-fetched or crap, but it's just more of, maybe not now, but in the future, it's a possibility that we may look forward to using this. I would say, in the geography, yes, we will use that every now and then, but we don't source a lot out of the US. So it is not something we are looking to go dive deep into right now. Even the ingredients that we have that come from out of the US, most of it is from a distribution center or something. So we just don't source outside. So I can't, as we grow as a company, it may be something we will look at in the future. What intrigued me more is the last case as the integration into our supplier management that I can see that being used to help us make those decisions that we need to make. And it's something, I mean, I know right now we're going through a whole process with our, I think a new one, a supply manager, and trying to figure out what can be integrated into it and how we can use it to the best of its ability. And this will be something that we can look at to see if it will help with that system. What else would I like to see? I know I've asked this question to several people with AI systems as far as what can be seen statewide? I know when we monitor FDA and USDA websites, a lot of their reports are usually nationwide for the US. And if there's something going on within a state that hasn't left its borders, it's very hard to get that information unless we are really, really Googling away. Or something, you know, as somebody just happened to mention about that outbreak or that recall, whatever. So that's something we have been looking for to get down to each state in the US. And also looking at things that are more of an enterprise risk. We had our pandemic that is lingering on from a couple of years ago. How does that affect transportation? How does that affect gas? How does that affect our supply chain? We didn't know it was gonna be as bad as it was, but to have that prediction in case something else was to happen, what is gonna happen to our ingredients and our suppliers if something like that was to happen again? And that's what I'm saying. That's interesting. What I hear you saying that you would like to see is first of all, to be able to be more granular in terms of geographies, to dig even deeper and look at even smaller geographies than of a country and especially in systems like the US. This is obviously relevant in your right. And I hear you also describing the need for something that will incorporate additional inputs or factors that may not be safety microbial or chemical related ones that do affect the safety of the supply chain and might lead into an emerging risk. That's very interesting. Dennis, what do you think? Okay, so if I can, I know there's a few notes there, but if I can just address it holistically, my feedback on this. So again, I'm with my colleague that there is nothing that isn't useful. It's how you use that information. So from a supplier base, from a supplier point of view, I expect to have a relationship with my suppliers. So I expect them to know what's going on. I expect them to advise me when they have a problem. I don't think I'm ever gonna pick it up with audits. So there's no point in increasing audits what I want is an ongoing monitoring process. So anything that enhances that for me, Nicos, is valuable. So if I've got some inputs coming into my internal system that tells me I should start to be more concerned about a particular supplier because they are vulnerable, because they are operating in a serial group that is becoming affected, then I can make contact with them to say, there is a problem, are you affected by that problem? And if you are affected, what are you doing to manage it? And if you are affected and you can't manage it, then I'll take responsibility for managing it. I'll increase my monitoring, et cetera, et cetera, for suppliers, for geographies, any reasonable business that has an holistic risk management process or crisis management process, should have predicted that some regions around the world are gonna have a crop problem or an infestation problem or an afrotoxin problem, wherever it may be. So geographies, I was sourcing materials from 160 countries, and we needed to be able to move from one region to another if we needed to. So early information of how to manage it, so early information about that is critical, not only to protect my consumers, but also very selfishly for competitive advantage. If I know there's a problem somewhere in the world, I want to get to the other source before my competitors do. So this is business value adding. It's not just about food safety, it's about remaining competitive. And when there is a global pandemic and everyone's affected by it, I want to be less affected than other people. The third bit, and again, as Takia said, if you can get information into your systems, so if you can enhance your own information with external information, that is always gonna be, again, be an advantage, and it's gonna allow us, what I call in my business, we call risk appetite. So every business has got a level of risk appetite it's prepared to take. And I think Takia said it in terms of the decisions of the VPs and the presidents, whether they are gonna do or not gonna do, or react or not react. That's based on the risk profile of a business. And if you want to take less risk, you have to add more cost into that equation. So anything that helps me to make that decision is critical. So they're all relevant, they're all useful, but they all need to be taken into the context of the capability of a business and what level of risk appetite it has for itself. If I were to, I think, seek more information, more energy from the systems that you can offer, I'm looking for the impact of more change, not just crop changes or supplier impacts. One of the biggest challenges I had was not the global pandemic, it was actually an election in Nigeria which changed the regime, which completely changed my ability to source sustainable palm oil from Indonesia. I had to then immediately start sourcing it directly from Nigeria. And I didn't have the checks and balances in place. So what impact would a regime change have on the supply chains? For me, it was a five million sterling change overnight, literally. The other thing I would like, I grow to be able to give me is specific data on material groups. So what's going on with nuts and seeds? What's going on with cocoa beans? What's going on with starches and sugars? With oils and fats? Tell me where the concerns are or the issues are around those. So I can look at maybe even reformulating, so I can take one material out and place it with something else. So this problem going on in a geography with a certain type of material, I don't even see it because I've been able to flex. And again, the biggest thing again is repeatable and predictable. My job is to make my supply chain repeatable and predictable. No cockups, no issues, no stops of production, no recalls, none of that. My job is to remove all of that risk or manage all of that risk. So anything that AI systems can do, I'll take AI intelligence all day long. I'll take any information that helps me to make a big decision, even if it takes my risk profile from 45 to 48, that's 3% better than my competitors. I'll have it. That's very interesting because what I hear you saying is that there is a spectrum of risk tolerance, let's say, which is a business decision at the end of the day because it has a cost and investment into putting everything in place and having people monitoring all these indicators around. And there is some balance and some trade-offs that have to do with where we will be positioning our business and our team. Because yes, we can have risk everywhere and the risk forecast everywhere and make sure that we react accordingly, yeah? Michalis, we have a wish list of things. So we got some good feedback on your use case scenarios and I didn't hear anyone complaining or saying that it's crap, I think. But I did hear a request for being much more granular in terms of geographies. I did hear a request for being able to incorporate other type of systemic factors or signals that might eventually lead to potential risk, such as a political event, elections coming up that brings some uncertainty and whether legislation will change and then it generates a significant amount of risk. What do you think? What can be done and what cannot be done with the technologies at hand? Yeah, first of all, Dennis and Akia, thank you very much for your insights and yours as well. And indeed, it's as you described, so as we initially said, all these AI models regardless of the insights that they're producing, they lie on the availability of data. So if there is, if there are such data out there that can help us make drill down further into the geographies and go on a state level, then that's a good thing. If there are not, however, this is a limitation of this kind of system in addition of such model. If no data is made available, no AI model can be trained to predict anything on the unknown without having some kind of indication of what could happen so that they can link and correlate it with. So I believe that going forward and how we could tackle this kind of limitation would be making more data available and always looking for more data out there, whether they're announced, whether we combine internal data with external data. This is the power that will help us train better AI models to be more detailed in their insights. Oh, I think what I hear you saying is that first of all, the models are as good as the data that we train them with, and that's very important. And one excellent example is the example of being more granular. If we can go even more specific in terms of location when we are collecting reports on issues and incidents, then yes, we can be more granular when we generate an expected forecast for a risk that will happen in this particular state or even be more granular hopefully. And the other thing that I hear you say is that in terms of the other types of signals, the more systemic ones that might lead into a potential risk, again, we have to find the data that will signal us, that will signal to us and the model that this might lead into a change in the system. So maybe it is a data source that is telling you about election dates in all the countries of the world. And there are some geographies where we will be putting some weights on whether this is contributing to higher risk or not. This kind of data signal is the type of data signal that you are describing, if I get it right, right? I can definitely do it. Okay, so you make it very easy even for me to understand. And thank you guys, but that's the million dollar question now, right? We have discussed about decisions that we can support. We have seen some use cases. You said that they're good enough and we have talked about some additional ones that Michalis says it could be developed, right? When it comes to the day that you will be taking the actual decision and you have to react upon something that the machine will tell you, what will it take for you to trust it? And which would be the real decisions that you would trust through the forecast that such an algorithm for machine will generate? Please give me a reflection about this. Thank you. Yes, so to have trust in the predictions, we will have to see trends over time and see how those come into fruition in real time. We will also use these, the data to be better collaborators with our suppliers, with other industry partners and basically share best practice and how we can do better because as we always say, food safety is not a competitive thing. We all want to make sure our food is safe. And we want to continue to keep the risks, the awareness of the risks throughout industry. We want to, we can use this to identify, help us to identify new testing, new methodologies that we need to do within our facilities to do better to manage those risks. And we can use this to better learn from others mistakes. Mistakes happen. We can say, hey, what did they do? What are we doing? How can we prevent that from happening in our facilities? In the end, AI will not be something we will use alone. It is something to help inform us and help educate us on what's going on and how we can implement it into our decision-making. And in food safety will always be, the decision will always be based on food safety and our business. So our business leaders will make that final decision with all the information that they're given, including information from AI technologies. Or it will inform planning actions and tools, but we shouldn't expect the AI to replace the food safety and business leaders one day. That's very important, what you're saying. Dennis? I completely agree, you know, I think to key is something perfectly, not really much more for me to add. I think there is a competitive advantage. I think food safety, like any other internal knowledge, is a competitive advantage. And if you do know more than your competitors, then you have an advantage in the marketplace. But I think this technology is useful in two phases, I suppose, for small to medium businesses, small to medium enterprises, it helps them to manage their risks better. So the additional information that sits within a small business that hasn't got global resource, that hasn't got its own laboratories, that hasn't got its own auditors, you know, it helps to manage their risk better for them. For global businesses, this helps them to manage, this helps them to better manage their procurement sourcing decisions. And this has a financial advantage because as I've said to you in the past, Nicos, you know, a big global business that is buying hundreds of thousands of tons of cereal or cocoa beans for chocolate or whatever, you know, it has two choices. It can buy based on contract and secure a price, or it can buy on the open market at a very much lower price, but that price fluctuates. That critical decision can be a 50 million pound decision one way or the other. So any information that helps a global business to mitigate that risk is useful information, even if it makes a three or 4% difference in your understanding of what you're gonna do. Am I gonna buy my cocoa beans on contract or am I gonna go on the open market or am I gonna balance the two? And what geographies am I gonna use? Am I gonna go for Columbia, Cote d'Ivoire, you know, Kenya, where am I gonna go? So this is useful information for that. And that also helps to manage, obviously buying sensibly, manages the integrity of the supply chain and protects the consumer in those instances. So it's different depending on what kind of business you are for me as well. But I hear you saying something very important and now I understand what you mean by competitive advantage is not that I would like to have access to this information or this focus for myself and let my competitor sell products that will get consumers, but what you're saying is that it will inform the other types of decisions that can help me be more competitive, like procurement then will increase my power, my negotiation power and procurement and sourcing decisions. Okay, let's see if we have a couple of questions that we can answer from the audience. I will stop sharing my experience so that we can take a look at questions. Do we have something there? So this is something interesting that they send me as a question, Paquia and Dennis for you. How would you feel if you were the company that was evaluated by the algorithm and highlighted as of increasing risk? Would this change your perspective? Paquia, what do you think? I don't think it would. I mean, we have been that company. We have been that company that I think every company has been that company at one point in time. So it's a conversation to be had. If we're the ones that there's an issue with one of our clients, then we're gonna work with that client to help fix the mitigate the risk and try to be, because we don't wanna lose that business. So we're gonna do what we can to make sure everything's good on our end and down the chain, down to our suppliers as far as what they're doing. So our clients can be, so they can trust us and make sure that we're doing what we're supposed to do. Then what do you think? Again, I agree completely with Paquia. From my point of view, some of my locations were audited 15 times a year and we didn't moan about that because not only was I manufacturing branded products, but I was also manufacturing for other people. So we manufactured under contract for businesses like Mars, Mondelez, and others. And the retailers like Costco, et cetera. So every time somebody came and saw our business or shared information with our business that was something we didn't know about, it's valuable. So for me, it's a valuable insight. So I completely agree. Any information shared with me about my business is useful to me if it helps me to remove or reduce my risks, definitely. So it's all about collaboratively being educated about risks, okay. So let's wrap it up. I will put this in the screen so that people can follow up and test a little bit. One of the things that Mihalis has presented, my last question to you as we are closing is if there is only one thing that you would like to highlight from today's webinar, the most important thing that you have realized or learned or would like to highlight to our audience, what would that be? Takia? I would like to highlight the importance of enhancing what you're already doing. You can always grow and be better and to increase the collaboration, have a partnership with your suppliers, that way things will go well with you and them as far as mitigating the risk. Thank you, Damis. Marginal gains. So in a cutthroat competitive world and as the UK has now left the European Union, we no longer have 27 associates, we have 27 competitors. So marginal gains for us, because I do genuinely believe that UK food is the safest in the world and how do we maintain that margin of marginal gain? Insight, AI, predicted technologies, I'll have all of them please. Okay, we have this. And from my side, I think I would keep this increasing understanding around how can AI and its insights be of help to the day-to-day food safety, the day-to-day work of food safety.