 Okay, so to be respectful of everybody's time, I think I'll now start the session formally. The session is being recorded, so you will be able to access it later after the session is over. But again, thank you, good morning, good afternoon, or good evening, depending on where you are in the world. My name is Neil Marshall and I'm here today to help moderate and present the session. We have a distinguished pair of guests with us. As we see on the screen now, we have Roger Hancock, the president and CEO of Recall Info Link and my good friend, Janis Stottis, the CTO and co-founder of Aguinal. We have our webinar today is to help us to talk about recalls, how to manage recalls and how to be better prepared for recalls. And we're trying to make this as interactive as possible and we're going to have really three main parts to the session. The first bit I'm going to share some of my perspectives from my role, my industry career and latterly with Coca-Cola and things to look out for and to consider as you prepare and try to avoid recalls. And unfortunately I went through several during my career. And then also we're going to have Roger talk to us more about how he could and his company can help you with mock recalls and how to also prevent recalls, but then also how to practice and prepare for them. And then finally we're going to have a more of an interactive case study from my good friend Janis. He's going to touch on the recent chocolate cases that we've seen around the world and then end with some Q&A open questions where you can post questions to us on the panel. And also we have some questions that maybe I maybe test Janis and Roger with to test if they've been asleep or awake during the conversation. Hopefully stay awake. Next slide please. So first of all, you know, as I said earlier there, I worked for Coke for 20 plus years and went through many crisis. And really this is just a few thoughts that I have and shared learnings that I've gained over that period. The first point really is about managing the crisis. One of the things that you need to really always do and can never do enough of is being prepared because you can't manage every element of that crisis. And as the second point shows with me grabbing my hair there and pulling it out, what's a little left of it. When chaos happens, it happens all around you and you really can't control it. And there's been some really tough times during my career where it feels like the world's come into an end. You can't do anything about it. You can't control it. You don't have all the facts, you don't have the data, but you've still got to react. And those are really difficult times to learn during your career. But it's also career shaping and personality shaping and it shapes how you learn for the future. And anyone who's managed through a crisis will know that they're the times when you learn the most in your career. And how to manage that brings you back to certain steps that you need to take and hopefully we'll share some of that through the webinar. But one of the big things, you know, it's a common phrase now, is how to pivot, how to adapt. Nobody knows all the answers. Nobody has all the data. There's always shades of gray where you can decide, do you know all the product that's affected? Can you recall everything? Can you pull it back from the retailer? What do you do? How do you decide on a course of action? But the worst thing you can ever do is end up, as I say there, in analysis paralysis and don't make any kind of decision. You have to make a decision even if you don't capture all the product. You need to do something. You just can't sit there because time is ticking all the time. So you always need to be quick to make a decision, hopefully make the right decision. But learn from it as well as an example, but it's all about how you respond, what are your learnings and being prepared. I can't overstress the point about being prepared in preparation and we'll come on to it as well. Practicing for an event helps you respond better when there is a crisis. So next slide, please. So some of these critical items to consider, you know, first point being internal versus external focus. One thing that you have to manage in the larger manufacturing companies is managing internal stakeholders. So that could be multiple functions. You know, you can have marketing, you can have legal, you can have operations, you can have finance, you can have the CEO and the COO and everybody else wanting to know what's happening and how you manage that internal communication if it's a multi-site, multi-country issue is critical. But you also have to manage the external communication as well. And so you've got two communication points to consider. One, managing internally, which can be different if it's a smaller company. But on the big ones, you've really got to know how to handle the communication and align all the stakeholders so that you're all singing on one hymn sheet. So you're saying the same message and communicating it and making sure people are aligned. Quite often, if you've got big countries and big corporations, you can have local countries sending messages out from communication teams that then are in conflict with the corporate or the global approach, which then is going to lead you to another problem. The next point about knowing what your role is, excuse me, go back where the slide stops. Don't move on, I've not finished yet. So knowing what your role is, everybody should have a role. You should have a team that's practiced. You need to do a simulation. But knowing the role and the role saw for crisis management and managing recalls is critical. You need to know who's doing what and you should have a team where the team's already defined before you get to the event so that you practice, prepare, and everybody knows how to respond and manage in the crisis. The next step, I said it before and I'll say again, training, practice and simulation are fantastic ways of ensuring you're prepared. You cannot beat being prepared by doing a simulation, a mock recall as Roger's going to talk about, so that you are prepared and you live through a simulation that should be as real as possible. Not only do customers demand it in most cases these days from the retailers and the big customers, but you also should do it for your own operational performance to make sure that you know how people would respond, what to do and to live as close as possible to the real-life event. The whole next point is about recall, planning and traceability. Not only the crisis management team, but you need to then connect all those pieces together so that you can identify the product, the ingredient, the material and isolate it, either withdraw it if it's left your control or it's gone to a warehouse or distribution center and it's gone to the retailer. That's the worst case scenario for a manufacturer, of course, because then you've got to pull it back as a public recall. But in some cases, you can do what's called a silent recall and you can extract that product without it getting out into the public domain and causing a media communication. But those, I guess, are less frequent and less obvious. And the last point about mitigation and containment is also important because the better traceability you have, the quicker you find the issue, the more chance you've got to contain in the problem and not explode in it even further. And the other point really around this piece is, this is generally, it will affect all products in all categories in that recall. So if company X has a recall, the associated products from other companies will likely be impacted. You see the baby food recall in the US now. It's a knock-on effect. So the manufacturers, if one company X has a recall, nearly always affects the others as well in one way or another. So yes, we should learn from these things and share as much as possible, obviously, and there's some limitations about what can be shared there. Next slide, please. And the final one, I think for me. So some of the tips that I've just probably said is, never think you're prepared enough because you've got to practice, practice, practice and train again and think where possible, what's the worst case scenario? Plan for the best case, but think about what's the worst thing that could happen in this scenario. And that should be built into your crisis management approach. Practice, practice again, I mentioned it already. Make sure that you have a designated person from each function in your crisis team, particularly for the bigger companies and bigger manufacturers, because you need to make sure the right functions are represented in those critical discussions when you decide in about, should we recall, can we keep it isolated, how do we move forward? And then the next thing, and also really, really important is making sure you have the data as soon as you can as clear as you can. Without the right data, you're gonna be making guesstimate decisions without facts. And as I say there, nearly always the recalls happen on a Friday afternoon or on a weekend when nobody's around and the whole hell's breaking lose as things start to go wrong. And most of all, most people will probably learn or come across a crisis recall situation during their career and they're better prepared for it. You are doing practicing and practicing makes perfect, as I say there and keep it as simple as possible because that's really, really important. So, enough of my tips and my sharing of my experience, I'd like to now pass over to Roger Hancock who is the CEO of Recall InfoLink. Sorry, just looking for my bio notes for Roger. Roger is a recall expert and president and CEO of Recall InfoLink. Before finding, found in Recall InfoLink in 2007, Roger spent 16 years at the retailer Albertsons as the director of food safety and QA, where he managed over 250 recalls. So, we really do have an expert here in Roger. And, he helped his company improve on those recalls, develop the processes and put the systems in place to manage those recalls before he founded his own company and move forward. So, we thank you to you Roger and over to you for the next session. Thank you, Neil. And thank you, Yannis and thank you, Agrino, for including me in this discussion. Neil, your comments were very insightful. I'm just gonna add one little piece. It was 250 recalls each year that we processed at Albertsons, not 250 over 16 years. So, lots of experience and we're processing thousands of recalls a year now for our clients. And what I would like to do in three quick slides is take you from where we are today to a transitioned or a modernized mock recall process and help you understand why we're talking about this, why this is an important concept today for us to consider. And what I wanna start is I wanna start with the goal of a recall. The goal of a recall, and I've done thousands of these, each one has been different. There haven't been any two that are identical, but the goal is always the same. Protect the brand, protect the consumer. That's always the goal that has to be in mind. And the path from deciding, okay, we need to do a recall, which Neil took you through some great analytics and practices to make that decision, but to get from there to a protected brand and a protected consumer takes action. And it's that action that really drives all of the processes between the decision and the protected consumer. So when we're thinking about mock recalls and we're thinking about a modernized recall, let's think a little bit about that action. Because despite the mock recalls that are taking place today, what we see, the data says, there still are crises happening literally around the globe on a regular basis when recalls happen. So the mock recalls today that happen, and I've been involved in hundreds and hundreds of these, really don't consider the end goal of consumer and brand protection. They're really more of a traceability exercise. What's the product that has a defect? Where does that product reside? If I can account for 98 or 99% of that product, I can count it as good. I've completed my mock recall exercise and I can go back to my regular job. What we need to think about is what action needs to happen where all of that product is residing. And what the data shows us over and over again, the data, I mean, this week the data shows us there's a gap that exists between what's practiced the mock recall as it exists today and what actually takes place and the pain that results in product being in distribution that isn't effectively recalled. And so if we transition and modernize that practice, to turn that practice into a comprehensive process, we can really change the crises that are happening again, for every company, the crisis is no daily or weekly thing. Although for some retailers, they get involved in a daily or weekly basis in their suppliers crisis. And we can turn that crisis into a normal business process. So let's go to the next slide and think further about how we can do that. Part of the reason that I'm involved in this conversation is because I can kind of tie together the mock recalls and the practice that Neil talked about and the predictive analytics and the artificial intelligence that Janice is going to talk about by saying, let's make mock recalls as close to real, as close to actual as we can. So if the data is saying, hey, there's contaminated sesame seeds out in the marketplace, maybe we use that as a mock recall scenario because sesame seeds are part of our ingredient statement or if we're concerned about sesame seeds and we've changed our recipe recently, but we didn't go through a package review, we changed our recipe, we didn't necessarily adjust our packaging. So maybe there's some ingredient statement that needs adjustment. Maybe that's a reason for a mock recall. Or perhaps we learned that our metal detector went offline and it's been, who knows how long and since the last test that we've done. So we can look at real life scenarios using predictive analytics and artificial intelligence to really stimulate what kind of mock recall we practice. And then of course, traceability is an important part of that process, but when you think again about protecting the brand and protecting the consumer, it doesn't end with traceability. So as Neil mentioned, identifying the internal players who need to be involved in a recall process. This is an important part of that mock recall, important part of that practice because when people practice, then they know kind of what's expected, they know the norms and they can even, you know, adapt a little bit or pivot as Neil also said for the actual case that they're in the middle of. But it's also important to find selected external partners who are willing to practice this process with you so that you can actually have the experience of saying, okay, this is what we're going to tell you if a recall occurs, we're gonna tell you what items, we're gonna tell you in this format, we're gonna give you instructions about how to handle affected items, how to dispose of them, et cetera. And this is what our expectation is of you when you are connected to us in a recall, we expect you to take these actions because we need you to help us protect our brand. We need you to help us protect the consumers. And so we're not only gonna identify some selected external partners for the process, but we're gonna follow through and we're gonna communicate with them. Well, this is what a recall would look like and this is what our expectations are because it's that gap that we need to close. So how do you build your practice to include that external communication while it comes from identifying some key selected external partners that are willing to practice this with you and then you can involve them in actually measuring, okay, how fast did we get the message out? How fast did the message get translated to the next level of supply? How fast were we able to protect our brand and protect our consumer? And then from those learnings, we can think about the rest of our supply chain partners and the processes we need to go through there to make our practice as close to reality as possible. So let's go to the next slide, please. So a modernized mock recall process really expands or takes to the next level this idea of mock recall. So whether you're doing a mock recall as an internal specified process or as a process to prepare for your next third party food safety audit or as a process to comply with your customer requirements, managing that mock recall in a modernized or expanded way is gonna be good for your business to protect your brand and protect your consumers if you think about it in a community way. It's not just what do we need to do, but what do our supply chain partners need to know and do what actions do they need to take to make sure that we're recall ready, not as only an individual company, but as a whole supply chain so that if something does go wrong, or should I say when something does go wrong, the recovery can be quick, the recovery can be a normal business process. The recovery can be comprehensive so that brand is protected. The amount of time that the recall is being executed is compressed and consumers are protected. And one way to again think a little bit further about this recall ready community is there's already connectedness across the supply chain for the purposes of sales, how can we leverage that connectedness for the purposes of product recovery when something goes wrong, when there's defective product that needs to be removed from sale. So involving that connectedness that already exists to leverage it to do a better job of removing product is another outcome that your modernized mock recall can have. And finally, in your mock recall process, think about modernizing the technology that you're using, using technology that is omnichannel technology because the way products move today isn't in a straight line from company A to company B to company C, but now it's omnichannel. It may go from company A to some distribution points. It may go from company A to some retail or points of sale points. It may go from company A direct online to consumers. It may go from company A to all three channels and what technology is available to help you saturate the supply chain with your messaging so that actions can be taken to protect your brand and protect your consumer. So with that, I'll transition now to Yanis. And let Yanis bring in how predictive analytics and artificial intelligence can actually do a better job of informing this mock recall process and help us prevent recalls. Thank you, Roger. I just got a question for you before I let you off the hook there. So apologies for getting the number of recalls wrong that you said, two 58 years. So that shows you really, really are a true expert. But how many mock recalls are you doing currently now with your company roughly? How many are you involved with? Hundreds a year. Okay. Are processed, yep. And mainly in the US or are you working globally or Canada, Latin America? We have an office globally. We have an international office in Europe. We're assisting retailers and manufacturers internationally. The majority of our business today is in the US. Okay. Thank you, Roger. We'll come back to you after. But now, thank you again for that. It's important now to use, as Roger mentioned, their technology and better manage the supply chains. And I think now Janice will now take us through a case study on recent salmonella recalls and give us some case study examples. As I mentioned earlier, Janice is the CTO and co-founder of Agroneu. He's also the brains and one of the brains behind the food archive platform. And he likes to refer to a quote we had from someone where the food archive platform is referred to as the Ferrari of the food intelligence solution. So I think he really likes that phrase. He probably likes his cars as well, but he likes to use that. So Janice, can you come to screen, come on camera and tell us how can AI really help us predict and prevent incidents from happening in the future? Hello. Thank you very much, Neil, Roger. It's great having you here. And it's great to have you all here. And thank you for participating. I will share our experience in putting AI in practice and more specifically, as Neil and Roger mentioned, how we can use AI to prevent recalls. So I will start first with a reference to the recent outbreak, the one with the Salmonella in chocolate products, which caused the recall of more than 180 different products distributed in over 110 countries. And this recall, such recall has already an estimated cost that exceeds this 60 million dollars. And this kind of recalls due to emerging key issues or known issues like Salmonella are more frequent in the last years. And I am saying that every time I would like to be at the position that I will not have something to share here. This is the goal, to minimize it. But still the cases that we have frequent such recalls in the last years, and these kinds of recalls are reducing the confidence of the consumers for our food supply chain. The specific recall also had a very important outbreak of a specific Salmonella stereotype that made more than 300 people to be sick, to become sick, and many of them were young children. So it's very critical, and we are discussing this with the community and the food industry that is very critical to prevent such risks and to protect the consumers, as Roger mentioned. So in the next slides, I will present how the predictive analytics would be used to help us prevent such cases. I will start with this slide that is showing which are the three critical scenarios in which the predictive analytics and the AI technology can add real value. So the first one is relevant to ingredients and the finished product's risk. By calculating the risk associated with all the ingredients that you are using, that a company is using to produce finished products, we can use in order to calculate this risk, we can use short-term, mid-term, and long-term. Predictions, the second type of scenarios has to do with the supplier's risk and with the facilities risk. In this scenario, we can use the ingredient risk and other key indicators to calculate aggregated risk scores and risk ranking to help us to identify which are the facilities and the suppliers at the high risk. And the third scenario focuses on the risk mitigation and the risk prevention has to do with the decisions that we need to make to become more proactive to activate preventive measures. So let's see all this in detail. But before going to the specific use case that I would like to share with you, I would like also to explain you how we are analyzing each decision-making scenario. So we are doing that by breaking down each decision in the critical business questions that we need to answer and to focus on the key question and to understand very well the question. This is our goal. So this starts, this methodology that we are using starts with selecting the right data. And after selecting the right data, for the parameters that it seems to affect the likelihood of a risk, to have a risk, a specific risk in my supply chain or supply chain. After that, we need to select also a prediction method that fits to the type of the question that we are trying to answer. So is this classification problem, is this a question of forecasting that has to do with forecasting? So based on the type of the question and the answer that we want to provide, we will also select the best prediction method. So let's go to the specific scenario. I would like to walk you through a real-life use case and the decisions within a manufacturer that has a variety that is producing a variety of chocolate products which includes ingredients like milk powder, cocoa, maize, palm oil, concentrated butter, buttermilk and other very important ingredients. And in the next slides, I would like to show you how such manufacturer can use AI to support the critical decisions for ingredient suppliers, facilities and finally to activate the preventive measures. So one of the very important things that such manufacturer of chocolate products can do using the technology is to be aware of the most recent risks and the increasing risks for his finish for products but also for all the ingredients in raw materials that he's using in the finished products. He is able to have a very good risk monitoring which helps him to identify early and emerging risk or an increase in risk. Like for instance, they increase by 200% of the salmonella risk in chocolate products which was highlighted very early by a system like Udakai. The other thing that he can also do that the manufacturer can do is to be very well informed about the emerging issues in the ingredients like the recent peanut solids and contamination in lecithin which is highly, which is one of the ingredients that is highly used in products like the chocolates. And this is a real thing. This is an example that I grabbed today from the system. So this is happening now. And the system can notify the manufacturer and can notify you about that early enough so you can activate and you can activate the preventive measures, all the checks or you can focus your mock recalls in this specific region, in this specific case of products. So the next thing that the manufacturer can do is that he can go deeper in the data and see how such data can be used to analyze the risk, a specific risk like the salmonella in finished products and using the data, using all the historical data of incidents, he can see that there were indeed several incidents in the past for salmonella in chocolate products with a peak of incidents in 2016 due to a large recall of milk chocolate products that were produced in the United States. So there were several cases already and known issue in the industry. So this is something that for sure we need to focus and we need also to get prepared well and to monitor the trends in order to identify something increasing that may affect our supply chain. So the next thing that he can do using coldest data is to study the very recent, if there are any recent incidents that can show us that there is an increasing trend and indeed for the chocolate products by starting the incidents from 2019 until 2021 until the end of 2021, it seems that there were several salmonella cases already in chocolate products having incidents like the one that we are presenting here in this slide. In addition to the incident, it's also very useful and very critical to monitor other types of information and to use other data like for instance to monitor the outbreaks data for salmonella cases that are announced across the globe. So you can also monitor the level of the outbreaks and not only the level of the final result which is the recall and the water rejections. So this kind of information, the disease outbreaks can provide an early signal for an issue like for instance, a salmonella outbreaks in food. Another thing that can provide a lot of value is to analyze the salmonella trends in incoming materials, ingredients that are used in finished products but also in finished products by using the large lab results data sets that are produced by the company, by the manufacturer or by the suppliers. So it's very important to follow, to monitor the trends of this data because there may be already an increasing trend there of samples being found to be positive for salmonella or exceeding a specific limit or just to be identified. So this is something that can provide a lot of value if we follow and we identify every and increasing trend that may indicate that there is an issue in my supply chain, my production line that I need to take into account or in the incoming materials. So I need to be careful not to add something in my supply chain that can contaminate the production line. So another very important part is the predictive analytics. Using our short-term forecast for chocolate product, we can see that already in October, 2021, our models were predicting that there is a likelihood and increased likelihood of the chocolate products being contaminated with salmonella. And this is actually several months before the outbreak that we are discussing now was very important that was realized in the market. So in the next slide, I am showing which is the time period from the point that we highlighted that there is an increasing risk trend for salmonella in chocolate product and the time that actually this outbreak happened. So it is six months earlier and this is a validated prediction and it was not only validated with a high prediction accuracy that we are continuously monitoring in our motors, but it was validated because a few months later, some months later, starting from December, but announced on March, early on March, there was the big issue of salmonella and as we are showing in the next slide, it was not only salmonella in the case of the products consumed by young children, but there were also other cases that originated from Israel with chocolate that was used in several other products. So this was a validated predictions and we were very happy about that, but not happy that it happened and they had so big impact in the market. Let's go to the level, to the other type, to the second type of decisions, to critical decisions which has to do with suppliers and the facilities. How we can monitor their risk and what is happening with my suppliers and the manufacturing facilities. So also in this case, we can use predictive analytics starting from very simple things like, for instance, checking or continuously monitoring which are the issues that are reported or the inspections that have been announced for the specific suppliers or for specific facilities, but also going using the predictive analytics as I'm showing in this slide that can help us to highlight which facility or which supplier will be affected because in this facility, we are using or we are producing the finished product that is at high risk or we are using an ingredient or raw material that is at high risk. And we can also, based on the predicted increase of risk in the materials or in finished products that are produced in this specific facility, the system can help us also and can highlight that the specific facility is at high risk. So we need to take action. We need to activate preventive measures. We need to get prepared for a recall, better for a recall. And this is the view for one facility, but the system can also provide, as I'm showing in the next slide, we are also able to have a live risk assessment in ranking for all our facilities or for all our suppliers and all our suppliers using internal factors that are estimated by the system using all the data that are published, but also external factors, sorry, but also the internal factors that a company may have. So if you are able, having such view, you are able to see which facilities are the highest risk without the need to manually estimate the risk for all facilities. And we can have there also as one of the factors, we can have the predicted trend for the risk in this specific supply or in this specific facility. So which are the lessons learned from this kind, from applying AI in order to identify which are the increasing risks, which finished products will be affected, which suppliers facilities and which are the preventive measures that we can activate. So in particular, we learned for the ingredients and finished products risks, we learned that the short term, the daily monitoring and the hazards analysis that we can perform using all the historical data was very useful for a manufacturer because it helped to check all the known hazards for the chocolate products and to check if someone else has previously affected the finished products or the raw material, the incoming ones that I'm using in the manufacturer facility. For the midterm, it was very important that using large datasets, the manufacturer can monitor which are the trends for Salmonella in the raw materials or in the finished products. Further, the predictions help also and are helping also to see if there's any other emerging issue that will involve in the next months. And in this way, using such a predictive analytics, the manufacturer can activate the extra preventive measures and can perform more recalls. In the case of the lessons learned about the facilities and the suppliers, what we learned is that using supplier monitoring but also live food safety profile for my suppliers and my facilities, we can activate and also consider alternative suppliers for the incoming materials or facilities or alternative facilities for the production of the specific finished products. And this kind of information for the midterm also, we learned that the manufacturer can combine not only external information, but also internal factors, information about internal factors and have and get a real time risk ranking for all the facilities and the suppliers. And this can help a manufacturer to adjust the lab testing by increasing the sampling frequency for specific raw materials or by adjusting the audits plan for the specific facilities and the suppliers. Regarding the preventive measures, what we learned about becoming more proactive is that knowing an increasing trend can really help in taking immediate mitigation action. So we can share internally the forecasting increase of the issues like salmonella and chocolate products and at all the different levels of an enterprise of a food enterprise and to take immediate mitigation actions. For the prevention planning, it helps a lot to make more frequent and dynamic the process of allocating prevented measures and budget to make more focus the investment that we are making in lab tests, in supplier audits and in facility audits. And the very important part is that using the predictive analytics and knowing which are the increasing risks, we can have mock records for the high-risk products and not selecting in a blind way where I will do the mock report. So we can be better prepared for a report by practicing on the high-risk products. So using the AI for the risk prevention may create significant business value. This is what we have seen by doing this kind of the business value exercise with several of the companies that we are working with. And this may create business value at the level of saving efforts for risk monitoring or risk assessment that is mainly done manually, but most of the values, of course, it comes if we put a preventive, a very good preventive approach in place if we can improve the already preventive measures that we have or activate them earlier so we can avoid the records. And this is where a very important amount of savings and the business value is coming when we have such a very good preventive approach. And we are doing this kind of business value assessment every time that we are working with a company so we know what year you can get back. And I will close with something that I always mention in my recent presentations that we all understand that there are many points in which AI can make a difference. So at the end of the day, it's not about, it's about improving the process, which is the transition that should be carefully managed. It's not an easy transition. As in the case of any other digitization project, we have to carefully consider how it will be introduced in the workflows that we have already without distracting the way that things are being done but the main goal should be to improve the things, the way that the things are done and to make sure that we are doing the best that we can. So a well thought of a planned approach is important and this will help the right people to be allocated and the appropriate routines to be put in place. So this is something that we need to keep in mind when we are talking about applying AI technology in the preventive approach that we have in general in the supply chain, in the food supply chain. So I will pass over to Neil. Thank you, sir, very much. Thank you, Janice. I can see, I just saw on that picture that we've used before as Mr. Elliott's name on the top. I must have missed that previously. I never noticed it. And also I have a different question for you. Do you have a Ferrari as you're so keen on comparing food archive in Ferrari? Do you drive a Ferrari? I think that I could drive, but I'm not driving it. The very important thing is to avoid buying a Ferrari and just have it parked outside the building because you don't drive it. So I don't have the Ferrari. That's a good analogy also for food archive, to make sure you utilize it correctly and use the information. So thanks for sharing, very interesting, useful information there around the dashboards and looking at different suppliers. I think just if we move on from the slide, before we continue to the Q&A with also with Roger and Janice, I've just been asked to share here. So we have an exclusive opportunity here for you today. If you want to sign up for the pilot of food archive, you can scan the QR code or go to the agronome.com webinar offer and sign up today and you'll get 50% off the pilot, which would be very useful for you and you can try and use it and gain some benefits. So the team will be very happy to hear from you if you want to sign up today. I think also now back to some questions and thank you Roger for going on. Maybe I'll just start with you. An easy question for you, I'm sure from your vast experience. But what's your greatest learning Roger or experience game from those many, many recalls that you've been involved with, either Albertsons or Sense, what's the greatest learning or a couple of learnings you take from that whole experience that you could share with others? Well, thanks Neil. Certainly the unavoidable learning one that you just couldn't miss is that every recall is different. So no matter what processes you have put in place, there's gonna have to be an application of those processes to the unique experience that each recall brings. Whether it's your first recall as a manufacturer because you haven't had one for 10 years and you have a whole new staff, you're gonna take your processes that you've tested and you've practiced and you're going to apply them to this unique event. If you're a distributor or retailer that receives a recall from one of your suppliers, you're gonna take your practices and you're going to apply them to this specific event because each event has its own unique components. One of the recalls that we're involved with helping our clients process just today off of the last week or so, you know, we started processing the recall for our clients about a week ago and yesterday a whole new item was introduced that was the result of further manufacturing of what is typically a retail product but that retail product got further manufactured into a second retail product and so an additional product had to be included in that recall and that learning took place over several days of the recall being practiced. And so, you know, whether it's an expanded recall because you have to add more items to it because your cleaning records don't support the limited scope that you've suggested or it's an additional product or it's an additional requirement for reporting that your recall insurance requires that you haven't added into your mock recall process and so you have to adapt there. You know, there's always something new. So being practiced is great because it allows those new things to not become their own crises but just an adaptation of what's already been practiced. So I would say the biggest learning, every recall is its own unique event. The practice prior allows that uniqueness of the event to be handled smoothly and in a way that actually ends up protecting the brand and protecting consumers because it's done efficiently and quickly. Kind of an unusual thing to learn that everything is different but it's obvious that everyone is different. It's obvious after the event and when you've got the learning, yeah, but it's never a wish at the time. Thank you for that. And maybe moving on to Yanis, I think there's a question from Steve Sklair in the chat as well. How long Yanis does it take after onboarding a client before you're able to effectively apply your predictive analytics to the company's specific situation? If you can quickly give a quick answer to that one. So it's, it does not take too much time. It depends on the complexity of the supply chain, of the number of the ingredients, the suppliers that we need to integrate or to customize the preferences for monitoring the risk. So it's very easy to do. So we are facilitating this process after importing all the list of the ingredients and the suppliers and the facilities that the manufacturer needs to monitor, he can get directly access to the predictive analytics for these ingredients and suppliers and facilities. So it does not take too much time after on average, after two weeks, after the onboarding process that we are applying, you can start using efficiently the predictive analytics. So the models are already there, are already capable to, and if we need to train any extra model, we will do so for some specific ingredients, but most of the models are already running and updated regularly. Okay, just quickly moving through then. Another question for you as well Yanis is how safe are these predictions? A lot of people always think about what's the confidence level in the AI? Is it really reliable? Can you trust the prediction? How is it really gonna help me? How much can I trust it? You wanna just comment on that? Yes, yes, of course. This is something that we discuss a lot with many people and we get this question. So it's important to know what is feasible with the prediction method and the data that we have at hand. So we are using a very good forecasting method that has been tested a lot with all the data, which can model with high accuracy, patterns like sessionality, periodic issues, the increasing and decreasing trends and can also identify anomalies in the data and the historical data that we have. But this method, this is a method that works very well for the forecasting and the method is not appropriate for classification problems. We are also, we are validating the models both from the accuracy and also from the use cases point of view from what really happened in the market. For the accuracy, we are using accuracy metrics and we estimate the error continuously to see if a model can perform well for specific ingredients and for specific hazards. And the other way that we are trying to deliver and very accurate prediction. The other way, the method that we are applying is that we are validating the models for historical issues, going back to all the cases that we know, like the ETO case, the Salmonella cases that we know already happened in the past for chocolate products. So it's very important to know the capabilities of the prediction method, as I mentioned in the first part of my answer. And it's important to monitor the accuracy of the method, but also to go back to historical cases and see and assess how well these models work. So when you apply such a good and analytical approach, then you can be safe that the predictions that you have can be very useful to predict and prevent recalls. Thank you. I think there's another question in the chat there. I don't know if Roger can also help with this. Do you think the question is from Mark, could you use predictive software, could that be something a retailer could demand from her producers? Is anyone seeing that yet or in practice? I don't think I have, but maybe Roger may have heard of something. It's a very tricky question to answer what retailers can require from their suppliers. So I'm not gonna say yes or no. I'm gonna say that having a relationship with your suppliers that produces continuous improvements to increase the reliability of the quality of the product produced and decrease the risk of problems is a best practice relationship between a retailer and their own brand suppliers. Exactly what processes to put in place or what requirements to put in place, that really is also between the relationship of the retailer and their suppliers. I think that there's on the one hand a growing trend to require best practices and also a growing trend to allow for flexibility in how those best practices are accomplished. And so again, we all know that there are certain retailers that are very prescriptive. This is the way you must do it or prescriptive. And there are other retailers that are more general. And so there's culture involved, there's relationship involved. But I think strengthening the relationship for the sake of increased quality and reduced risk is a strong case that can be made. I like your answer, Roger. Good answer, thank you very much. I think maybe as we're already over the time, I'll just ask one final question just for you, Yanis, just to put you on the spot. So the ROI report that you presented, what size of company and product range does that really refer to? And can a company get a similar report before signing up for your deal, your offer that you mentioned? Yeah, thanks for asking. This is a good point. And the case that I presented is for large and very large companies. And it refers to a case with over 1,000 ingredients and more than 500 suppliers. So this is regarding the first part of the question. Regarding the second part of the question, a good estimation of the ROI needs some data from the actual use of such a platform. And that's why it is, we are usually estimating it at the end of the pilot. So however, we can have an initial estimation before starting the pilot if you provide the characteristics of your supply chain and you provide us the number of the ingredients, the number of the suppliers of the facilities that you would like to monitor and for which we would like to predict risks. OK, well, thank you for the answer, Yanis. Thank you, Roger, as well, for your time today. I think as we're over the time and we should probably conclude. But thanks, everybody. I think that was a very interesting session today. And remember to sign up for the offer on the screen or go to the link if you'd like to participate in the 50% pilot. Thank you very much. Thanks, Roger. Thanks, Yanis. It's been a pleasure. Thank you. Thank you very much. It's a great pleasure. Thank you so much. Thank you. Bye.