How AI-Powered Kitchens Are Tackling Food Waste at Scale

I hear more and more is people start asking questions about how AI can help me. For the back of house operation, the first wave was more like phone dial in ordering and so which is okay, it's a kind of natural evolution application. But those still are not the most fundamental. And I think the most fundamental is going to be that people now looking at connecting the dots realizing the POS data is not the best data for them to actually manage their inventory and everything else. And then food waste, they realize it because now so many case study come out, people appreciate yes, you can actually make a serious impact. You're listening to Food Tech Talk Supply Chain Insights from Farm to Fork, a podcast discussing the latest trends and technologies in the food and supplements industries. Brought to you by TrustWell, featuring conversations with regulatory experts, quality and safety champions and thought leaders across the industry. I'm your host Katie Jones, CEO at TrustWell. If you're loving the conversation, don't forget to subscribe and stay tuned to the end of the episode where you'll hear our guests tell us their key food industry predictions for the next 12 months. All right, let's jump right in on today's episode of Food Tech Talk Supply Chain Insights from Farm to Fork. We're about talking joined by Fengman Gong, CEO and co founder of Metafood X. A serial entrepreneur with a background spanning AI, security technologies and practical innovation, Fengman has now turned his focus to one of the world's biggest challenges, food waste and inefficiency across the food supply chain. Through Metafood X, he is helping kitchens use simple AI tools to reduce overproduction, improve planning, lower costs, very important these days and create more sustainable operations. Today we'll explore whether AI can truly transform the food system, where the biggest opportunities lie and how technology can support a more human centered future for food. Welcome Thingman. Thank you so much. Great pleasure to join you. Katie. So what first inspired you? You have a career spanning cybersecurity and transportation. So what first inspired you to move and shift career into the food industry? Thank you for great question. My career has been up to that point before starting has been very rewarding, a lot of impact and also made a lot of friends and learned a lot and you know, reluctantly also growing older during that whole journey. But then indeed right one of the things is solving the IT problem versus then my last startup actually brought me to the transportation sector ride sharing although it's bordering both helping them with the cybersecurity and also driving safety. So that was transition for me to get a taste of how much impact when you get closer to everyday life of people? I mean we were dealing talking about like 30, 35 million rides every day. The platform actually connects. So that was initially what impressed. And then after that one of the things is we always when I grow up or even today. Right. We always think of the necessities for people that is food, housing and transportation, clothing, those other things. Then of course food absolutely come on the top. To me that that was the point if I was going to do something, anything close to a startup in Denver that I'm going to look for something that is really potentially impactful. So that's of initial thing get me to be interested in looking at the space. It's a supply chain that affects everyone in the world. You know, we're deeply passionate about food safety here. And my oldest son has a life threatening food allergy and I believe that no one should be afraid to eat especially fresh food and have access to fresh food. And so it is certainly there's lots of areas in the food industry to center on and to do some really great work. So that's awesome. Yeah. And the interesting thing is initially right interested in food. But actually indeed when I started was thinking says okay, one of the things as us individual right. Then thinking about how to help people to eat healthy and eat responsibly. That was the initial actually intent. And we did some research. But then quickly we realized that actually with you know like this 40 billion annual US loss and waste on the supply chain, we actually realized that the consumer side of the help is very important. But the timing was not right. It's hard to solve that problem. So we actually end up quickly deciding we have to focus on helping the business and have a B2B solution to start with. And we actually do see how this data we collect. Right. Accumulate would help down the road with consumer. I love that. I think it's a great model especially for software if you want to have a meaningful impact on the supply chain. Right. There's a lot of consumer demand for sure. But going straight to I guess the source makes a lot of sense. You've described food as one of the most human centric businesses. I love that. Why does that perspective matter when you're designing software and a technology solution? It's a number of things, just a number of examples. Right. To illustrate this, of course being human centric, one of the things actually was I learned early on in cybersecurity. That's where what we realized is regardless what we are trying to do, various technology in the Middle trying to optimize, increase accuracy coverage. But we realize there is always human behind it and there is always human getting impacted. But then coming to food actually made that more pronounced. Because just take one example, right? When I first look at this and worked went to some of the future food summit. One of the first thing I realized actually I criticized some of the people event that time. Everything was so much focused on purely either it's basically alternative protein or plant based which is great. But my feeling that time was we were kind of more influenced maybe by investor by the tech trend, the hoopla than the real food problem. Because today we know when it comes to the food, I mean the flavor is important whatever you do because people have to take it, have to eat it, right? So that was one example, right? Food pronounced that human centric aspect. That is for whoever consumes this product, which is food, it got to be human, otherwise there is no point. And then the second aspect is along the way as I was developing this solution, I actually learned a great deal because we went to a number of food service operations right from college and university to including food and beverage for the basically the corporate basically food operation group. And we quickly learned how human connection is in that environment. Because when you find out people working there, they actually have been working there for many years and they actually try to learn help each other. Of course part of that is also because you are preparing food, the product is consumed by human. But then there is also a lot of human labor and team behind it. So in the end it's still then human who are producing the food will be the ultimate gatekeeper in terms of as you mentioned earlier, the allergy or it's simply just nutrition. Today more and more we begin to realize how much what we eat is connected to how we live the life and the health. So that becomes kind of more intertwined. I mean you, you have to involve the human to make that decision. And they were the ultimate like a decision maker or the gatekeeper for the consumption that was. I think it's just amplified for the food. As we were working on it, I realized it more and more looking to improve your supply chain. At TrustWell, we've been connecting the dots between food industry and data for over 40 years to give you more control and visibility. Our comprehensive food logic platform sets a new standard for compliance, transparency and quality in the food industry. Request a free demo. The link is in the show notes. So let's talk about the wave of AI that is coming across every industry and impacting certainly a significant amount among technology companies. So balancing that human centered approach, a human centric need in food with AI from your perspective, how does AI create meaningful impact across the supply chain while still maintaining that human centricity? There are a couple of points we can discuss. One interesting thing is first of all, we realize that the food service in particular, let's say if you take the university of dining hall, indeed that atmosphere is still very important because one of the very experienced food service leader from UMass, Massachusetts can tune their customer when we work with them, with the team, his team, he just showed us, I mean how once a while they have those. Either it's a supervisor or Cantu himself because a lot of their people eat in the dining hall with some of the faculty, with the students. And they just walk around, actually sit down, just ask the students, how is it, what is it, right? And do they see anything interesting or something not right? So that is one aspect of that human touch for service. And I think that's why we have a strong service, right Industry. To me, that cannot be simply replaced by machine. People talk about robots. So that's one aspect long term we do. We can discuss about that vision, right? What that looks like. There is one other aspect is we need to look at AI as a powerful, powerful tool, powerful assistant. The assistant could be very autonomous. That is okay, but it's still helping human. I think this is the fundamental point. But if you look at today, right, if you view AI for us, for everyday life, it's best to view them as an intelligent machine. That intelligent machine aspect is very critical because as an intelligent machine, it's very good at certain things. It's going to surpass human. We know they can be working any hours. They don't have to take the kind of stop we take. They are very good at the heavy lifting, both physically, literally and also number crunching. Because we are great, we can figure out something, but if we are overwhelmed with huge number of things, that's where we just cannot keep up. So that is the kind of understanding to see that not to be kind of influenced or just blindly says okay, you know, it's cool because we are able to chat. It seems to be so smart, just like be able to carry a conversation. And that's where that's the difference. We have to go beyond that. When you come to the food service, then the first thing probably universally is actually able to help the operators to gain visibility on what's going on in operation. I mean this of course all the way carries from the food kitchen up the supply chain. That was very interesting because when we get into this we realized people do have a number of very meaningful and very sound best practice they have accumulated, right? Such as trying to do do small batch when you have to feed a lot of people and try to do just in time and all those things. And of course a manual planning. That's probably one of the oldest wisdom. Everyone knows it. But then if you look at the reality, not everyone is practicing this to the same degree and is not producing the same results. And if you look at that, then that quickly bring us to the issue. Do they have the data? First of all, is the data granular enough, useful enough? And then there is another aspect of it. How do you get this data without burning out the staff? Because we quickly, when we started using the product, we realize it because people know they want to do best practice. But it's the tool, it's the environment, it's a thousand other priority that is preventing them from implementing that best practice. So that's how we see exactly how the AI and the future robot should come in to actually take those heavy lifting that human is not good at and then create that system that can be like coexist to serve the human. So that's how we have been approaching this and we feel we are on the right track, we are rewarded with the results. Well, what I love the most about that is it is still very customer centric approach to AI where you're really listening to the customer and understanding exactly what their need is as opposed to developing AI just for AI sake. And I think we see that a lot in technology where there's just, it's the buzzworthiness of AI whereas it's not actually practical. And you emphasize that, that there needs to be practical solutions, not just in technology, but you know, I think an AI application as well. So what makes an AI product truly usable in the food industry from your perspective? So we start with where is the kind of the heavy lifting that is bogging the system down. And then we need to come up with a solution that would get adopted into the environment, fit into the workflow and then produce the data or the insights, whatever you call it. But it's something people see it and know what to do or what to change and they'll get results. So I would illustrate this one with a few points. One thing is we are talking about actually getting the understanding of the food consumption. So this is a tough one because it almost like at the very end of the data, even up the supply chain, we should Be relying on this data. If we have it right, we know how to translate. So one of the things if you look at today, people asked to track what they are producing, how much gas consumed at the end of the day. And then what they end up doing is they come up with various ideas. Initially, some places not very busy, they may still try to use a scale. And then they would record, they would actually weigh. And then for the scale they still have to tear and they have to write down pretty much it's the pen and clipboard for people to do it. And this is where when any reasonable environment very busy quickly people cannot keep up then what people start doing. And it was interesting. There are some environment then have the creativity. So it's okay, let's just. At the end of the day, let's count how many different pens we went through. So what that means is you actually have to remember what pen was holding, what. And not to mention, right, you actually now trade off the accuracy. So there is so much of this, right. And the people who even go into from like a institution down into the restaurant, they also have that issue there. People know exactly POS told them the sales unit, how much money and then the cashier get. And then they look at their monthly bill, what they bought. And that's how they have been tracking, try to see how their margin is. But they also tell us in the middle how the foods get managed, get trimmed, get produced. So what's the yield? It actually. So they don't know the actual food that people ate and how that translates the whole thing and the loss. And in that process in waste. They know it's a big part of their margin is going there, but they just don't know how much. So that was the first thing when we have to kind of understand, right, where if you actually solve that problem, then everything actually follows to give them the data. And then they can make very quick adjustment steps. Once we actually have menu item by menu item, day in day out, how it is consumed. So they know both from a forecast point of view and. And from the operations point of view, if there is an issue in operation. So that was one aspect starting with solving that problem. And then you need really make it so that you give people the useful data. But you should not be increasing new complexities for people. This is where in particular, if you think of the staff day in the day, our staff is the one. It's the business. They have so many things to take care of. So then one of the things is how do you make this Solution assistant tool so that the smaller, the less the interaction you require of the staff, the better. So one example for what we end up doing was through multiple iterations, we see how it's creating impact. So we want to get many item by item kind of data. But then we have to create this all in one multisensor integrated solution, which we call an AI scanner. What you can think of is you have a platform which have an embedded scale, but it's integrated with the 3D camera on the top with infrared sensor on the top. So then everything, when you put the food on the platform, it automatically recognize what it is automatically take the net weight and the temperature. Everything goes already to the cloud automatically. So that was. We make that workflow, we call it like a touchless. And all of a sudden including the different serving pans and containers. Any container in that environment, you just need to register it. Once that's it. It recognize the container and you can have any combo right with the food. So that is example what that solves is how easy to get the data without adding the extra burden to the staff. And in this particular case, we also realized it actually did something more because what end up happening is when this device get inserted into the workflow, it actually help them to form a habit. It's simple. And even for the new staff. And they removed some of the onboarding of the staff training. They had it passed because you reduce the complexity, right? They have to know first do this, then do that. So those are the examples. And we really need to make it so that it's solving a specific problem. But it just naturally blend into. It's almost like helping them, but it's not visible, not necessarily. They have to talk to for it to understand something. The approach makes me think of another podcast. I guess we had. We had the youngest master chef ever actually on the show and talked about how R and D and you know, menu innovation should watch the food waste, essentially. So watch what's coming back from the kitchen. So it's interesting, you know, to understand, right, what's getting wasted. Which, you know, there could be a variety of reasons for that, but how R and D may be able to then loop that back into menu planning. Because, you know, like, let's say, you know, because consumer trends, they're forever shifting, but ultimately at the end of the day, like what? You know, let's say the students in the university, right, what they're not eating. Why would you continue to invest in that? So that's really, that's really interesting and just A better way. It's funny that we see this a lot with our customers as well in terms of tracking quality issues that if it perhaps just wasn't getting done before, there isn't a status quo, there isn't an established process. And then you start to see that process and you know, we'll see also working it into manager bonuses, things like that. Right. You start to see it really start to pick up and it becomes a part of the culture then of the company because you've made it easy, you've created ownership over it as well. And that's really interesting. That's great. I love that Chef comments Because the device actually can work in one of the mode through software configuration post consumer mode. What end up happening is actually if you deploy the device as a dish return and you only need a sample of the student or people who ate, not only you would be able to get the record of what is most left on the plate. We also give them the option they can say okay, the food today is great, it's okay, or just I don't like it. So that's something, right? The kitchen, the shelf, they only need to look at the data, they can look at the rating, they can look at what the plate look like and that would actually give them a good idea of what could be happening. So indeed I think this is the chef and the staff, they are actually very creative and once we expose to them, show them what the system the tool can do and they actually start getting ideas of what the other things they want to do. I love it. That's awesome. So when you think about where companies can start. So what are the whether from a technology adoption AI but specifically on the topic of food waste, what are the low hanging fruit opportunities today for whether it's restaurants, food service institutions, other types of operators, where can they start? Yeah, I mean I think this is where depending on the environment they may have a slightly different issue. The majority of the environment where anyone who needs to feed a large number of people from any of the dining setting up on two kitchen actually they would have this critical issue is to simply get the consumption data. See the interesting thing is consumption data is influenced of course by the population. Right. You are serving and then it could be depending on at school it could be affected by weather, time of the day, day of the week. But one of other critical thing that carries across is also your manual lineup. Because in that environment whenever you give people the choice, it's very interesting. It's a very important factor because once you have the Manual lineup. It affects how people are going to take different food items. So that's why in those environment the simplest way is to start. Because you can deploy at any food venue, right station you choose. And then once you deploy it, it simply become a day in and day out the food going to the service line. Every pan gets scanned. Actually it's about one less than two seconds. Everything is locked. And then when you have anything left in the pan, you can scan it back. It would know because it recognizes it attracts what get consumer what get leftover. At the end of the day they have the notion of the data not only for the menu. How preferred, how people are taking relatively the manual lineup. Most consumed versus least. And also what was overproduced. And that's the two data basically will give them immediate action. They can do whatever the way they want. They can optimize their manual lineup and switch out the items. Or they can basically adjust the amount they are going to produce. So that is in any of the group dining environment. And they will be amazed to see how much they will be able to save. And that's one thing we have so many case study. Now we went further. So we know if you take a catering environment. Then we actually added the features for them to track in terms of beo. So for a given kitchen you have so many events several dozen. But you can track everyone individually so that it would know which event. Because catering I think people are coming around right before they know customer maybe already paid for given the lineup. So they start from a kind of bonus and everything. They thought we're not too worried about it. But then now they know if they can save a dollar, that is a dollar of their profit. So they're coming around. We make it easy for them. But then when you come to of course the restaurant it's slightly different. Because each environment have a different. Because the restaurant in particular for any like a quick service. It's the same actually same benefit directly carried. We have a customer not only use this for the tracking, right? Knowing the pattern day in, day out. So help them plan. But also we support them. One of the feature. Because they care about the quality, how fresh the food is out there. For the when the customer come to get the food, you serve them. So then what we have one of the feature is actually a small feature for the system. But the customer find it useful. Is for every menu item. When they cook the food and prep it before they put out. When they do the scan, they can configure a timer. The timer actually give Them automatic reminders. If this one you said 30 minutes out there and you should be replacing it and the time racks you give them when it's one minute to go and things like that. So they find it really useful. When you have so many things people need to do other things. This is where the machine although it's a trivial task but for human impact. Right. It actually is a lot. So I would say for sure. I mean this any environment they just need to get that very accurate granular consumption and everything they will find it flow from that with an immediate impact. Starting step one and then building from there. That's really interesting. The timer on when to you know to refresh. I guess there are food safety applications there as well. Right. So knowing you know when to you need to put it down. Right. You cannot be leaving it out there and things like that. Yes, absolutely. Absolutely. Well, we've got two questions left before we close this out. This is a question we ask a lot of our. Our guests on the podcast. What's your food industry prediction for the next 12 months? For the 12 months. This is in terms of the technology or in terms of the. I guess I can only speak from a technology point of view. Sure, yeah, let's do the technology. Yes. I mean I feel people are really now one of the things I hear more and more is people start saying okay. People start asking questions about how AI can help me for the back of house operation. So I think that means they are not making the connection between what AI can do and what they need before that. And the first wave was more like. And people started doing some, you know, let's see auto right phone dial and ordering. And so which is okay, it's a kind of natural evolution application. But those still are not the most fundamental. And I think the most fundamental is going to be that. Right. That's the first step people now looking at connecting the dots realizing the POS data is not the best data for them to actually manage their inventory and everything else. And then food waste. They realize because now so many case study come out. We really people appreciate yes you can actually make a serious impact. So that's the first one. It's going to be. We expect it's going to be very, very quick kind of adoption on that. But then the other interesting thing is where the rest of the AI would evolve to help. We have seen talked to a number of people also as potential partners that include one is you probably know about this cooking robots. So people are learning because there are only a few cooking robots have the potential. And I think their application would expand. But only to the degree there is a combination of use case. That is a small number of menu items requiring huge amount. Right. For any of those catering environment. In combination with solving a couple of other problems. They have talked to me, for instance, one thing is still prep. Prep of the ingredients is still largely very manual. And even feeding the right amount of ingredients into the cooking robots as you imagine it can totally destroy a recipe. And that is still a problem. So they actually seriously talk to us. So we know there is going to be someone have to create a reasonable end to end solution for specialty restaurants. So I'm going to expect to see more of that coming. And then in that process, of course the data end up still right smack in the middle. Because it's a data that are going to drive that whole process for them to be efficient. So those are the two areas. The other areas, one thing is kind of more a pessimistic aspect. That is the kind of things purely based on interaction, some kind of extreme customization for a given customer. In those restaurants, I think it's still hard. It's hard to imagine people's taste, people's variation. So that in theory recommendation is okay. But when you come to fulfillment, I think then they'll run into this fundamental problem. Yeah, it breaks down. Yeah. The highly personalized nutrition field is very interesting. Based on an individual's particular whether it's preference or you know, genetics. Right. What do they need to thrive. But you're right. At scale it becomes difficult to do today. I mean I definitely see that, you know, I see that happening. I hear some of the requests right from for example elder living or different assisted living environment. Because it's very critical I think for those environment. You definitely have the volume for that environment. Right. It's niche. But you would need to build an end to end solution. So you can streamline it. Absolutely. Because to me that is the big potential. And related to that one is almost like maybe we can take a slice of even supply chain. Because we know at the end if you have a group of food operators, if they have the accurate data based on where they are, who they are serving, I can see that data. You build a system, go up the supply chain to help the food providers. Because today they're doing their forecast based on the buyers putting in orders. But then they know they change their orders. So that is interesting. I think that is at a macro level. I definitely would hope to see people actually see that coming. Because the impact is going to Be huge. At the consumer level, there is a different kind of extreme customization, as you just mentioned. Absolutely. And then at the bigger supply chain, if you take a slice, it's probably also making a big impact on the people up the supply chain. Absolutely. Well, we'll wrap up with this last question here. You're clearly incredibly passionate about the food industry and making that pivot from cybersecurity to food. When you look back years from now, what success story from Metafood X would make you most proud? Yeah. And I think we can look at this from maybe two angles. Right. One angle being an entrepreneur, very passionate, always trying to solve new problem. And I think I'm glad because we have that belief. We see basically a potential for technology. Right. Technology to come in to make an impact. And so it's that belief also believing the technology and the human aspect and those insights drive us to say, okay, it's something we can do. And then I'm glad. Then we end up proving it can be done and in actually producing the impact. So we still yet scale. But that is something I think is rewarding in the sense when you try to do innovation, it's that determination. And then when you believe it, then as people say. Right. You just have to stick to it. Although it could be hard. Right. Because not everyone would understand it. So that's one aspect I think it's rewarding with that vision we pursue and then we actually be able to produce this product, get used by more and more environment. Right. And then if you look at if I was going to step back, I think this is connecting into the bigger pure impact. Because now we indeed we find it rewarding because we solve the problem for the food operators to actually cut down the overproduction and the waste. So anything associated with it we actually preventing that food waste, preventing the cost associated with energy and human labor. So it's actually more than just recycling. So that is direct impact to the operators. We are hoping that we see the kind of the indication, early indication. We are breaking that myth. People was talking about why food service is always going to be backbreaking and you are going to be always settled with low margin. We just feel we are not at the cusp of actually insuring people. You can do so much better. And that is one directly impacting the food service industry. And then the second one is about the climate. Because this gets into when we started because I always felt if you think of our food supply is so critical, you always have a supply and demand to manage. And early a lot of effort were focused on the new generation, new supply. It's expensive, it's long term, it's valuable, but it's long ways to go and this is where the low hanging fruit come in. I view this, what we do, solving the food issue as kind of getting low hanging fruit for the climate and for the sustainability in long term. So that is also very rewarding I think. Again, you know, take us, take the team. I really appreciate the whole ecosystem, the partners to actually see this, to believe in it, to support us to this point. But I would say as you asked about earlier about the consumer, I do believe of course the next big impact is going to be something really easy to use, right? Because for consumer you have to cover their intake 360 days, 65 days a year for it to be meaningful. So that means the solution need to be very easy to use and very convenient to be able to do that. And if we get to that, then I think we will close the ecosystem between the consumption and the production and that will be wonderful. I've been at this, the cross section of food and tech predominantly in traceability and transparency and all of this for about 10 years now. I completely feel your call to the industry as well as you just have to keep plugging along because change does take time and I'm very glad that you are a champion in the food industry and towards food waste, sustainability and improving climate change. So Fengman, thank you so much for the perspectives and for sharing your story and about what drives you in the food industry. I really appreciate you being on the podcast. Thank you so much. I enjoy the conversation so much. Thank you for joining us for another bite sized nugget of food tech talk supply chain insights from farm to fork. Make sure to subscribe and if you love the podcast, leave us a review on Apple Podcasts, YouTube, Music, Spotify or wherever you get your podcast to learn more about Traffic TrustWell and request your free demo of our technology platform that connects food formulation products, nutrition analysis and compliant labeling with traceability, recall, readiness and supply chain transparency. Please Visit us at www.trustwell.com and click Get Started or click the link in the Show Notes to learn more. Thanks for listening.