Drones Are Changing the Future of Farming

Speaker A: Consumers are expecting more transparency in what's happening in the food. For the industry, it's an obligation and no longer will it be good enough to just think about one step downstream and one step rapidly. Hitting the minimum isn't going to be good enough.

Speaker B: 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 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 joined by Jim Trohanas, Chief Technology Officer at Fruitist. Jim leads the development of AI, computer vision and predictive analytics systems that are transforming how fresh produce is grown, harvested and delivered to consumers. His work is helping Fruitus move beyond traditional field sampling by measuring the quality of individual berries in real time, enabling smarter harvest decisions and greater consistency across the supply chain. So today we'll explore how practical AI is reshaping agriculture, improving food quality, reducing waste, and creating a more data driven future for fresh produce. Welcome to the podcast, Jim.

Speaker A: Thank you for having me.

Speaker B: So you have spent years leading large scale technology transformations at McKinsey and now you're working at a, uh, produce company. And so Frutus sits at the intersection of this between agriculture and technology. So I have to ask, what was it about this opportunity that made you think I have to be a part of this?

Speaker A: It's a good question. I mean, outside of being absolutely obsessed with our product and knowing that we have the world's greatest barriers, I think the draw here for me as a technologist in ag is seeing a material opportunity to make real tech work in an operating condition, that it's right for that change or for that next step in the evolution. For me, having prior experience and exposure with Fruitus as a client of mine in the past, and knowing what the opportunity, what the potential looked like, not just from a, hey, this is an interesting problem to go build a team around and deploy, but from seeing the underpinnings of what data really exists within the organization. And I think that's a key differentiator for us. Layer on top an incredibly entrepreneurial and driven leadership cohort that is very, very understanding and very pro technology in terms of how we serve the next generation of consumers out there with healthy nutritious snacking. We know that technology is going to be a vehicle for that. Based on admission alone. I was compelled to join and I think we have a fantastic opportunity that we've been exploiting with all the data we have in really interesting ways.

Speaker B: First of all, very cool and I'm a huge fan of big bold pivots in careers because I think it's where you grow. And personally a huge fan of giant jumbo blueberries for the record, just putting that out there. And I think the quality especially I think in the berry category can be quite variable. And you see this, that most growers are relying on sampling to assess crop quality. And when you think about how the work that you're doing is really reshaping the connection of quality and growing practices and other things like that, why is that approach of this kind of more old school sampling approach, why is that approach no longer enough for modern agriculture and arguably the modern consumer?

Speaker A: It's safe to say that a practice that's been around that long of kind of the approach of understanding how crops performance is evolving in the field, the consistency of results are there, right? Like if you are just a regular commodity player and we say hey, we grow blueberries and we're shelving a non premium product, you say okay fine, that might be sufficient. In reality, there's plenty of opportunity there. Layer on top. Again, the need for us to really see ourselves not just as a commodity, but rather a premium product and a consumer product CPG rather than a berry producer. We have an obligation to provide a level of transparency, our operators to result in a consuming experience for our end customers that is above a very high bar. In order to do that, we can't go measure two or three bushes per hectares or whatever it might be and make a thumb in the air a guess or which way is the wind blowing in terms of quality outputs. You just go deploy labors to go pick and then whatever berries we get is kind of what we have. So we have to be much deeper in the data and much more understanding at a certain volume of what's happening on the field for the counter bushes per hectare so that we can make really good decisions when we deploy our folks to go pick. It's really the cornerstone of us being able to deliver a consistency and quality that's unmatched. It requires us to have a much larger sample size than just three bushes per hectare. Three bushes per 100 hectares. Um, really depends on who's operating the field.

Speaker B: Looking to improve your supply chain? At ah 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 FoodLogic 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. And you're doing this using computer vision. First of all, can you, for our listeners who may not be as familiar with what that means, can you provide an explanation? How are you using computer vision to manage quality? But then also how do you see that leading to enabling, I guess the growers to make different or better harvest decisions?

Speaker A: It's a very technical solution to be honest with you. So rather than defining based on a technical definition, why don't I just paint a picture for some of the listeners? So what we have currently running are autonomous drones equipped with cameras that are able to record video, not just images but video. And then we use AI in the background to be able to process that video, uh, and identify pinpoint individual barriers on the bush. And not only are we picking up which barrier, but we're tracing that barrier for frames. There's a significantly hard problem when it comes to deploying computer vision in the way that we are doing and that you have occlusions with leaves that are covering berries. If you take one image and then the next you can't really differentiate if you're double counting or not. So computer vision is a big picture, is just a machine's ability to identify things within a frame. Our approach to computer vision is video based and traceability based. So we're actually looking at the movement of various through each frame. It's a rather distinct piece of technology that we have exclusivity around and it's something that we're very, very, very excited about seeing coming at scale in the future for ourselves. So computer vision is the ah, ability for a machine, a computer to be able to identify as if a human eye would using optics, quality characteristics, sizes counts. It is a very advanced piece of technology that we're using. It is a multi domain area of how do we say, you know, identify the berry, be able to run the sizes, be able to look at colors at pixel level for an image that's being taken pretty far away from a bush. It's a hard problem but we're using drones in the fields to do this. And what we're solving for is both an increase in the volume of observations we have in the field. So we have confidence behind the sample size that we use for judging when things are getting done or when berries are producing at a level that we want to be able to pick and meeting that high bar, but also accelerating autonomy. We have a issue here of you can't hire enough people. It takes 30 minutes to potentially count to, uh, 2,000 and push. We don't have the labor force needed to count the quantity of Aries in the field to arrive at conclusive. We'll say samples of what's actually happening in the field. And that yield number that we're trying to solve for is essentially the most important thing that drives every planning decision we can make, operationally and commercially, for that matter. And it's also a component to some of the other optimization models we have, which are around quality. So it's not just about predicting how many or how much is coming from a certain field. It's about understanding what quality we're actually seeing from the berries within that field. Um, and quality obviously has many characteristics tied to it. It's not just ripeness. It could be brix and acidity and other elements.

Speaker B: So you have the ability to measure the ripeness, size, and then any potential, you know, issues. Right. So quality issues or even just defects. Right. In a product across the entire field, what impact do you feel that that has on, first of all, your level of visibility, on the just the food quality and consistency. But then how does that empower your team, even maybe your sales team, right, to give better information to your, either your consumer or your retail partners?

Speaker A: Like I mentioned, that visibility is really the most important driver behind having great trusting relationships with our retailers and also a consumer experience that brings our customers back for more. A very positive, delightful eating experience. Red sweetness, red mild acidity. It's all rooted in our ability to deliver to our customers and those retailers with an understanding of what they're going to get. So we're eliminating what we call this berry roulette, which you may have heard about. It's a common term around the office. For us, it's really a sticking point. We are interested in playing a guessing game. We really want make sure that we can rise to the level of expectation for consumers as a premium product. It's all starting from what's happening in the field and having visibility at a volume, at a scale that's big enough for us to have confidence in what we're packing.

Speaker B: When you think about rolling this out and as you're gathering this data and understanding, right, all of the inputs that then Yield certain outputs from a quality perspective or consistency. What surprised you most once you started gathering all of this information and having really data on every single berry, arguably.

Speaker A: So there's a couple things. One, we had a hypothesis that we were missing critical bits of volumes of information and we very scientific approach here, test an experiment. We found that the outcome can be influenced quite a bit by increasing the volume of data and having that level of visibility. So it's kind of shocking to see how much control we can actually deploy into the operations through some of these analytics and the visibility and the traceability that we're trying to institute in all of our fields. I mean we're still early in that journey. We have drones live and a good portion of our base operations, but we have ample opportunity to grow that program and see the fruit kind of come to bear as a result of that effort. So for me, the extent of which we were able to enact a change or see the value come in terms of quality improvements. Consistency is huge. And the other thing is providing the visibility to the business. It was very, very interesting to see how well our business teams were suited in being able to act on that information. You can imagine everything from labor planning to pack house planning to the way we engage with our customers, all of that is, it just piles up. It's like one layer of benefit and impact on top of another one. So it was very refreshing to see people just grasp the data and see the work that takes place internally in our operated teams.

Speaker B: There's been varying levels of tech adoption across the entire food chain from the tractor just at a base level, but all the way to now. We're seeing some really interesting enhancements and information that we can gather around quality. Right. And then how that connects through to um, essentially the ultimate objective, which is to sell a really great product. But it's still an industry that's fairly somewhat archaic at times. From a tech adoption standpoint. I think we see a lot of great ideas, but then when it really the rubber hits the road in terms of how to actually implement. So what were the biggest challenges in making computer vision work in just, you know, real world farming?

Speaker A: Well, as you can imagine, we deal with elements we can't control. The weather. Uh, that's kind of a problem that we all share in the agriculture space. We absolutely cannot control the weather. But fortunately enough for us, behind the solutions lie a very talented team and years of decades of observations. And as we think about the advancements in the tech ecosystem outside of agriculture and some of the synthetic data generation There are levers that we can pull to increase or rapidly increase the volume of data that we can use, whether synthetic or real observations to improve some of the base models that we have now. And those are going to be levers that we pull and we continue to pull in the future to just outperform years past. So for us, whether it's hey, rain kind of gets in the way or cloud cover gets in the way with lighting, these are things that the industry is also looking to solve. So we have great partners and a talented team to do it. But yeah, we're going to continue to turn on it. But here's the biggest takeaway. Regardless of those moments of time, that creates a challenge. Challenge from a technology perspective, simply the scale and the outcomes that we're seeing from the programs that we're running and the technology that we're deploying will always without a doubt outperform a three bush per hectare or a three bush sampling technique. Right. So we're not going to let perfection be that and be a progress here. We're going to continue to push the issue and that's really where our focus is. It's continuously improvement on, um, all of our models and all of our AI capabilities. And if you look at a naive kind of counterpart or a naive baseline, you're still going to do much better.

Speaker B: You talked a little bit about this earlier and I loved your use of the term CPG that you see your product as a CPG brand. Because I have also, I would say, counseled our teams to think about produce marketers essentially as just that. Um, and I think especially as consumers are looking for more fiber, more fresh options, but the number of variables around produce compared to something that's on a manufacturing line. And you have a tremendous amount of control, right. In terms of what issues still happen, obviously, but from an operational standpoint. So when you think about, I find it fascinating you could having that complete picture of quality and yield to then translate that into your supply and pricing and all sorts of things. Like from your perspective, the tighter control that it seems that you have on that kind of quote, manufacturing process really growing, where have you seen the biggest operational gains from having it seems like just a much more tighter control in

Speaker A: terms of the quality it affects entire value chain for us. And as a company that's vertically integrated, right. We see the impacts from bush all the way to distribution and to our retailers. So I mean we're tracing impacts through and through. You know, on a farm operations level, you could think about just simply being able to Move a high volume of workers in the field to go perform a harvest operation, to go provide agricultural ag inputs in the fields, to manage irrigation. All of this is based on our ability to really pinpoint with a level of precision what's required when and what's coming down the line. It takes time to move product from a field to a pack house. It takes time to stock that pack house, it takes time to move the labor. So it would be easier for me to say, uh, actually, I don't even know if I could pinpoint something in our business of like, what isn't impacted by our ability to control that process or to have a higher level of predictability for what's coming in terms of volume and size and then also what qualities, uh, will exist in the harvest that we're seeing. It's a hard question. Everything touches it, everything is based on it.

Speaker B: Yeah. And you have to think that not just obviously consistent experience for the, the end consumer, but being predictably reliable to your retail partners. And it sounds like even maybe perhaps being able to, if there is an issue, then communicating that well in advance having that information. So AI is, you can't go a day out of hearing about it or what's the latest. And we're still, from our perspective, I mean, over the last couple of years, seeing a significant amount of adoption and, and more embracing, I think, than what we saw a few years ago with more fear. But from your perspective, how do you see AI helping producers identify risks before products reach consumers? And as we're recording this, we're in the heat of a cyclospora outbreak and with a, um, you know, very difficult time trying to understand where it is and uh, it seems like multiple. So from your perspective, when you think just across the entire industry, how do you see AI helping progress things in a positive way moving forward?

Speaker A: So, one, we're very fortunate that it has not affected us whatsoever. Two, I think the acceleration of insights is going to be a critical driver to shifting the identification of resolution left, right, fresh produce. But regardless of what you do, potential exposure is always a risk. There's categories, there's dedicated teams. We have an entire department dedicated to this that uh, has incredible amount of oversight and governance to make sure all testing is done earlier and often, but shifting that visibility left. And it's very much part of the work that we're doing. You're just providing a higher level of resolution for what's happening in the field and a higher level of predictability because then we can point our resources in certain Areas such as testing which are applicable to the next harvest or what have you. So the more we could shift left and have more predictability of what's happening, the better off we'll be in being able to more quickly identify what's happening in the field and resolve it before any consumers are ever impact. Introducing additional layers of technology through the entire value chain. So not just focus on what's happening upstream, but introducing data traceability throughout the process allows us to then both identify and resolve at each step of the journey for some of these barriers. And obviously working with our retail customers and our partners is a critical component of that as well. Everybody has an obligation to our consumers. We do our part, we expect our partners to do theirs as well. And we actually govern that quite a bit for those that we work with. So for me, AI will be the accelerant and it will be the thing that allows us to get to higher volumes of testing and more accurate, precise diagnosis events in the future.

Speaker B: Very exciting. You know, when you think about over the next five years, what, uh, technologies do you believe will have the biggest impact on the AG sector in the future?

Speaker A: Well, some of the work that we're doing in computer vision in our space, incredibly challenging as a result of the size blueberries and the kind of phenological characteristics of them. Uh, whether you have occlusions because of leafs or other things, it's a, uh, challenge. But I think that is going to be the, not just an unlock for how we get that consistency for ourselves, but also in the industry, but also the foundational layer required for us to move into, dare I say, is more autonomy robotics in the field or in our pack houses. We need the ability to inform physical AI and how to behave and how to interact with the physical world. And that requires an optical element to it, right? Like being able to see and navigate, um, and identify. And I think that's going to be something that those on the bleeding edge are going to continue to push and see the rapid development. And then those who are lingering behind, I think the uh. And there's always going to be those two flavors, those early movers and those late adopters. I think seeing a lot of this impact materialize in case studies that get served in the industry or through financial performance ultimately of the companies that do invest in these things, it will only pull some of those lagging companies forward and say, look, if not now, then when are we going to invest in building our data foundation, in building the capabilities internally to partner with our business teams to do this? It's an eventuality. I'm predicting a lot of movement in the next 12 to 24 months. But those would be my bets I think autonomy and then just a continued investment in ag. As you mentioned, there are some ladders in the data element will be a critical thing in the next 24 months.

Speaker B: Yes, we work predominantly in supply chain and it can be quite alarming at times to see how paper based some of parts of the supply chain chains still are and that you know, getting to faster decisions again especially around traceability and being able to respond. But then how do you actually then use that data to essentially measure and contracts between two companies where you're guaranteed a certain amount of freshness and you can actually capture that when you understand when you've received that right. When was that, if it's produce, when was it packed and being able to just raise the visibility of that information. Because right now there can be some, you know, some black holes essentially within the industry or within the food chain and oftentimes that's where that information gets lost. And then we're dealing with what we're dealing with today with some of these outbreaks.

Speaker A: It's going to have to change. Consumers are expecting more transparency in what's happening in the food for the industry. It's an obligation and no longer will it be good enough to just think about one step downstream and one step backward. Like you have to be able to go all the way through. You know my thinking is that's my aspiration. Hitting the minimum isn't going to be good enough. Like we have to extend beyond that point. And I think data and technology analytics are going to be a must have in order to get there. It's not going to be a optional investment in order for you to satisfy the needs that's going to come. You're definitely going to have to invest in the infrastructure and the capabilities to make that reality for your company and ultimately to satisfy the consumers.

Speaker B: Well done. And where can our listeners get their hands on some of these amazing berries? Where can people get and buy Frutus?

Speaker A: Today Union, the sor will likely appear on the show. So Whole Foods, Costco, Walmart, you're going to come across us. And um, I encourage our consumers to put what I'm saying to the test. Pick up a pack, enjoy those berries fresh, pop them in your mouth, go pick up another 2, 3, 4. You're going to be hooked after that. And let us know if you're seeing the consistency. But that's what we're after for each berry you eat, you should have a consistency of a delightful experience and a crunch that can't be met.

Speaker B: Well, we ask everyone the same question as we wrap up our interviews. What's your food industry prediction for the next 12 months?

Speaker A: I think we're going to see a lot of success stories in the AI space. And the reason I say that is the barrier to entry in some of the most advanced technologies historically and some of the capabilities that were only available to those like OEMs, John Deere's of the world that have vast engineering teams, et cetera, those are now going to be within arm's reach of companies that can do a lot, very low hanging fruit for them to think about how they can use and leverage data in more interesting ways and use some of these models. So I'm excited to see the application of whether it's Genai or more advanced AI technologies like the computer vision, how that is going to play a role in the development of the industry. Not just us. We know what our strategy is and we know what our vision is. We're chasing that. But I think the industry as a whole is also going to hopefully, maybe I'm an optimist, adopt a lot of that.

Speaker B: I love that. Access to technology to spur innovation and not uh, just having it available to the big players. That's great. Excellent. Jim. Thank you so much for the insights. I am now craving blueberry pancakes, so I know I'll be getting on my instacart to see where I can get some uh, some fruit. And it's been really great to talk with you, um, and to hear about the really innovative and incredible things that you all are doing and all in the name of consumer safety and trust. So thank you so much Jim.

Speaker A: Thank you for having me. It's always a pleasure to talk about. And for the berries on the pancakes, keep them on top, don't cook them inside, they're too good to be cooked.

Speaker B: Okay, excellent tip. Thank you Jim.

Speaker A: My portion.

Speaker B: 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 podcasts. To learn more about 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.