Replicating the Farmer’s Eye w/ Kia Behnia & Mason Earles, Scout

Hey listeners, I want to tell you a little bit about Offset, a proudly independent commerce platform and brand studio based in wine country. I've personally used their ecommerce platform, Offset Commerce for over a decade, and I've found it to be one of the most intuitive and functional platforms on the market. Don't just take my word for it, establish luxury brands and leading independent producers and merchants like Brian Estate, Bedrock Wine Company, Arnold Roberts, and Acbee Fine Wines use Offset Commerce and Love it. Go to OffsetCommerce.com to learn more. That's OFF-SET-Commerce.com to learn more. Welcome to X-Shadow. The podcast that navigates the business of wine with unique perspectives and insights with your host Robert Vernick and Peter Young. Welcome to this episode of X-Shadow. Today we're going to be talking with Kia Benias, CEO of Scout and Mason Earl's, the CTO of Scout. This is going to be a rough-yup series on Vity Vity Tech with AI in the vineyards. Can you please give me and Peter a brief overview of your backgrounds? Maybe we can start with you, Mason, since you're the first time on the podcast. Yeah, well thank you guys for having me. I'm super excited to chat with you about what we're doing at Scout and also in my other position, which is UC Davis. Yeah, I'm a professor at UC Davis and I am the co-founder and CTO of Scout. Today I'm going to be mostly wearing my Scout hat, but if I am mentioning anything about UC Davis or my position there, I'll be sure to kind of call it out. I've really spent the last decade working on AI, applications and food systems for the last seven or eight years. A lot of this is just like working at Apple prior to this. I did a lot on plant physiology and my PhD and then Scout started from that intersection of agriculture, AI and really like a lack of the technology that could make a big difference in the vineyards. That's kind of bit about my background. And Kia, you were on episode 150 where you talked about some art farm and you gave a lot of information about your background, but Kia, what part of your background led you to Scout? I think as I mentioned before, my wife and I own a vineyard and a wine brand, you know, tempo is the wine brand and Kia, Tra vineyards is our vineyard. We had embraced technology since the outset. We planted the vineyard 14 years ago. I had the pleasure to work with three great vineyard management companies during those 14 years. And what I found is that unlike a lot of other businesses, there wasn't a lot of data that was being used for decision making. And that kind of inspired me. My last company was swank. The tagline was data to everything company. We would take operational data and different verticals and help decision makers make fast decisions. And then we brought AI operational intelligence in. So we kind of knew firsthand that for our industry to prosper in the digital age, it needs better solutions, better technology solutions. We have the people. We have experts. We have really amazing expertise that right now is happening. We just don't have the data and the pace of change. The challenges we have with climate and climate chaos, the labor cost and the budget pressures all lead to more technology rather than less. And maybe before we dive in, a quick clarification is the company called Scout or Ag Scout. We just had a prevene of Monarch Tractor on and he called it Ag Scout. I'm more familiar with it as Scout, which I think some of our other guests have actually mentioned it, you know, as Scout before. And so we just want to make people understand or clarify what is the name. So the name is agricultural Scout is our official name, but the DBA is Scout. So we go by Scout, AgScout.ai is our website. So that's where people can find us. We don't mind getting called either as long as it's in the context of agriculture and not sports scouting or other things. Although many of the concepts apply. I mean, again, we keep finding that that term scouting means a lot and it's a it really captures what we need to do more of, which is look ahead and scouting what's going to happen ahead of us as opposed to scouting backwards. So now that that's sorted, can you give us a brief overview of Scout and when and why it was founded? Yeah, I'll kind of start then, Kea, you're feel free to add some details in as well. So like I mentioned a minute ago, we started Scout. Really, I mean, the origins would be the first few weeks of when I started at UC Davis, we had GoPro's out in the vineyard attached to ATVs and just started recording lots and lots of data. From that point, we started building up hardware and trying to kind of building kind of complex the PCB boards for the electronics and all this stuff and ended up doing that for several years. We're going to commercial growers and kind of internally with researchers at the university. And eventually, we found it Scout about 2022. This is when, as actually a funny story, I don't know, Kea, you're planning on telling this or not, but I'm going to mention it, which is that so Kea and I met at the so you see Davis, we teach a program, a wine executive course. And that wine exact course brings in wine exacts from across the world who are interested in kind of running wine business. And so Kea and I met there for the first time. Kea was there kind of the probably most interested and clearly most knowledgeable person in all the tech related questions we had sessions we had. So we got to know each other then and then he kind of jumped on. I started reaching out to Kea for advising and eventually was able to talk him into joining us, the co-founder, to kick us off the ground really. So I just wanted to shout out to both that wine. exact program is an amazing thing to do, but it also really was one of the origins of Scout. Yeah, and one of the things I loved about the exact program just to put another plug in is it had a track around business and it had to track around viticulture and all of G and you would pick which day you go into. And it was amazing because that's when I learned how complex the budgeting aspect and forecasting aspects of the business was while on the operation and all side it was hard. The one thing I remember asking Mason is what's the best sensor? This was my question here in wine exact. What's the best sensor you've seen in the vineyard? And he said it's the farmer's eye. And really that sparked the idea around what if we could build a digital version of the farmer's eye with all the abilities to remember exactly where you saw a problem. And this is really kind of goes into the genesis of how we went from being a camera and hardware centric company to becoming an intelligence company that we are today. We're all about data insights and intelligence. Yes, there's hardware that's involved, but that hardware is now standard commercial off the shelf. And that allows us to also be agnostic and partner with folks like Monarch and robots and various other things that are coming down. I got a funny piece added that too, which is that I think I remember Keo when we were talking about this and we had been running the first set of pilots where we had been running around with these very inter complex camera system that we had set up and we were talking to Keo and he said, look, we've either got to be a hardware or a software company, but can't be both. And if we're into software, clearly we're focusing on the AI piece. And so that was sort of the moment where we said, well, how would we do that? And we said, well, let's center everything around the phone where we can literally plug cameras into the phone. We can communicate via the GPS through a phone. Everything that phone becomes the central processor because again, I think the thing you said to me was there's one thing we know that's going to happen. It's that phone technology is going to keep getting better. And it's going to continue to improve. And so that was something that we, conscientious decision we made right then for the next season coming up and that was never looked back. So that's kind of our goal is, hey, we want to use the hardware and do a good job with the best hardware. God, let's let the scale, the power of scale take care of that and we'll focus on AI, video culture and how to make a grant me supported by AI. Well, it's a good segue to understanding really what is Scout's solution for vineyards because our first question wise, is it a hardware as a software as a combination? So what is the system? So let's talk about it. Let me use some everyday examples. When you look at Zoom, when you look at podcasts, there's a piece of software that actually delivers the value, the content. And then it needs to integrate with a variety of hardware. And the reason podcasts have taken off is that you don't have a proprietary piece of hardware that you have to listen to podcasts. This podcast can be listened to in my car, you know, via appels, downloads, stores, Spotify, any of the podcast platforms that it goes in. And that's why it works. So we can close the ecosystems that require you to purchase the entire stack from one vendor. They're very difficult to build. Second of all, they really limit the value of existing investments people have made. If they've already bought a camera, if they already have a smartphone, if they invested in RTK GPS, why should they buy a different one just for scouting? So our strategy has been to build a broad ecosystem of recommended hardware that we test and then use standard non-propied theory systems so we can kind of expand that. And that's the ethos that runs through the company really over the last three years. So we support Monarch as our preferred electric tractor. We work greatly with them to take advantage of their row follow and lot of features that they bring in. But we also work with a 20-year-old Kubota tractor that somebody may have bought and is not ready to switch out. Now the Kubota tractor does not have any cameras. So we offer a set of standard cameras that you can basically attach their very low cost less than $300. Their waterproof. They've been battle-dested. But this allows you to retrofit existing equipment and not have to buy a brand new device to get the value of scouting. We've taken the same approach exactly. Smartphone is the brain for our product for data collection and smart phones are getting cheaper, faster, more AI capable and they're easily available. So if somebody drops one or rides over one, they can run to the shop that day and keeps counting. And so our software, just to be more specific around the value proposition, what is the capability that Scout provides? We really have focused on four key capabilities that literally move the needle in every vineyard of any brand of any size. Number one, automated vine counts, automated inventory and mapping. This is an age old problem. We bring in interns oftentimes and young enthusiastic workforce and put them on the absolute warcing that nobody loves, which is go out there and count a bunch of vines and use a clicker and write it on a sheet of paper, then manually input that on a spreadsheet. We automate that entire process. The camera can be much better at understanding what's in the ground and applying those rules. So we not only count, but we classify. Is this productive, nonproductive? Is it grafted or young? What's the trump diameter? How many shoots does it have? So we do this extremely well four times faster than the human on foot, because we can ride on a tractor that's already going through that row or drive a TV at nine miles an hour. Second use case is around crop performance, which includes in our world both vigor and fruit. And our goal is to help you measure. So if you're doing regenerative tests versus conventional, due to a B test would quantify data. If you're looking for where my underperforming vines have that data that can instantly show you what vines in a block are underperforming, not all of the plants are doing well. So this has massive, massive opportunity in this tough economic climate where nobody's interested in buying and expanding vineyards to get more out of the vineyards you have with the farming costs that you're already spending. Again, you're paying to farm those acres. This is about how do we farm it smarter? How do you get more out of the investments you're making? Third, and this I can't wait for Mason to share more with you is literally the hardest problem we've come across in vineyards and vine culture, which is yield forecasting. We've met with over 200 viticultures and organizations. Nobody seems to be confident in how they do yield forecasting, especially when they need it. This determines whether you're a buyer or seller of fruit in a given year. Lot of inefficiencies in our industry exist because we don't have forecasting. Honestly, I can't figure out another industry that doesn't have forecasting. You don't manufacture cars without knowing what your forecast is on how many cars you give them. While everybody's talking about doom and gloom, I think this is the area that we need to focus on is how do we fix things that are broken in our industry that we can fix, hopefully using AI and technology. Last but not least, health performance and virus mapping. The second biggest threat that we have, especially in Napa and Sonoma, I hated when I drive around in Silverado or Highway 29 and I see a full block this being pulled back. I wonder what could that block have been saved? Had people become more vigilant and had the right data, could they have tested everything, etc. So we use AI for virus detection. We use the rest to scout to map these virus vines and run analysis on what is the strategy that you should take. So those are the four, sorry for that detailed answer, but there's a lot of meat behind each of those. Kia, you mentioned that the industry doesn't do forecasting today, but it does. There's yield estimation. We sample and then we extrapolate based on that sample. It may not be at that accurate, but it does happen. This may be a more accurate solution, right? So just Kia and Mason just kind of like re-understand back to the original ethos. It is the farmer walking the vineyard their eyes or is the best tool that Mason had commented on. So just drawing analogies to that. So basically cameras are the eyes. Some form of tractor is the legs kind of walking through the vineyards. And then the brain, the processing is the CPU on the smartphone that eventually, or and the scout cloud, right? Okay. And then there's just some kind of network connection as well. So those are kind of the requirements for it needed. So you need the visual. Even on the network. Yeah, that's a great point. And even on the network connection side, we've been working a lot on offline. So that when it comes back into network synchronization, we sync up. So it's not that you need it at the moment. You just need to eventually have it to get the synchronization back to the cloud. So how scout works is that we take 20 photos of each plant. Those photos are taken while an ATV or a tractor is driving up and down. So we can do it at nine miles an hour. Also the GPS point of each of those plants is capture. And then the data, whether you're in connection or after you're connected gets uploaded to our cloud, where that data gets processed into three items. Counts, measurements and patterns. And those counts, measurements and patterns results in a set of insights that are actionable. And those insights could be around crop performance around inventory and asset management. They could be around disease detection or disease risk. They could go and input into your yield forecasting model. That's also AI based model that does use predictive analytics to forecast what the yield might be. So in terms of adaptation or people have picked up and using scout, I'm assuming the problems in NAPN's number could very greatly from other parts of the world. So can you tell us a little bit about like the reach of scout and how many customers you have in what different parts of the world? Yeah, absolutely. So today we're a US only company. However, we have customers that are multinational with the full intent of piling this into us and then using it overseas. We'll be expanding the company internationally next year as our business continues to expand. Today we have somewhere between 50 and 100 customers managing over 300 blocks. Over two million vines are under management today. And we have analyzed and processed over 56 million photos. Just two weekends ago, we processed more images than we did two years ago for the entire year. So we've been on an explosive growth path right now. And we have a number of global brands that are using on us. Many of them started with 10 acres and now have expanded to 100 acres. We are in NAPA, very active, Sonoma, very active, Paso and Santa Barbara are next and we have our anchor customers. Our customers fall into three categories, winering, slasher state, their veneer management companies. And the third is real estate investors or veneered investors, both buyers and sellers that use scout as a way of doing due diligence on a property or setting up remote monitoring for the absentee owners or the groups that own the property outside. So what then are the core benefits of the solution? Is it cost reduction, quality improvement, environment, labor? Can you articulate what people get in terms of a financial return? So I think in terms of financial return, again, when you're capturing data four times faster and 10 times more data, obviously there's a cost saving, but that's not where the benefits stop. There is efficiency that's gained. There's a lot of productivity gains out of the existing teams. What this means is teams can manage many more acres. So the ratio of viticulture is to acres managed becomes much, much better. Quite frankly, we give people their weekends back. A lot of these viticultures are overworked. We met with someone who said, I walk eight hours a day every day and I still can't get to all the sites that I need to get to. We've had windmakers that spend hours driving from NAPA to Sierra Foothills and back when they could be actually in front of their computer and remote accessing the data and having the data come to them instead of them having to go to the site. The value of yield forecasting, again, is massive, particularly around early season forecasting. Mason talked a little bit about how forecasting estimation are different. We really don't like the word estimation. I think that's what the industry's tried to do for a while and that's part of reason we keep missing. We've had customers who have gotten their forecasts wrong by 40%. So in 23 vintage, just think about vintage of the century, there were people with great fruit with no home. How many of would have wished we knew exactly what that forecast was and had the right tank space for that vintage? It also came back for you. You may think you have a yield, you can underestimate it and then all of a sudden late season, you're a buyer. So we've worked with customers like Gloria. Gloria Ferrer is a customer in sparkling, very difficult problem around yield estimation. We're working with them very closely and what we've been able to do is give them early forecast and being able to kind of have some predictability within the business. We have other folks that have been working with us to really quantify the benefit of knowing what's coming out of their estate and then how should they formulate their contracts in terms of the fruit that they're buying. The last part is disease. Disease has a massive downside because I think the last report showed that over $17,000 per acre of damage that a virus blocked. you have to replant brinks. That's not a one season thing. That's a three to five year kind of set back. And when you look at the cost of testing, it's about $40 a test if you want to run a PCR test. Most people don't budget to test every plant every year. So the idea becomes how do you maintain virus levels below 15%? There's some great research from Monica Cooper and Sarah McDonald that came out where they have best practices that said you have to do vine by vine management under 15% but once it gets to 15%, you may lose the whole block. So a lot of far solutions around how do you keep the longevity of that vineyard not just within a year but for generation to come? On the yield forecasting part, how early are you talking about where you have an accurate sense of what you're going to get? Because obviously, Mother Nature comes into play at some point and changes throughout the year, which should I would imagine changed the forecast. Yeah, yeah, I think what you're pointing at right now is actually the reason we like the word forecast as opposed to an estimate. And whenever we say estimate, I don't think there's a truly technical difference in the definition, but I think that as people were used to dealing with forecasts in a number of daily aspects of our life, right? Whether that's the weather forecast, traffic forecast, there are other things that suggest exactly what you're saying. Peter, which is that like, hey, there's different sets of information, which have already we've already seen and we know versus pieces in the future that we don't know, but we have other info that tells us about the likelihood of those things that are going to happen in the future, right? So we try to break down yield into a pretty, I think, well understood, but fine grained equation right into the yield equation. And that really boils down to, hey, all right, let's at the beginning of the season, we know that early on, we're already setting a maximum yield ceiling based on the pruning that we're doing and what happened the season before, right? And so each of these steps, so we start there, we start moving into saying, all right, now the plants start growing at Budburst, we're going to end up starting to understand how many shoots emerge. So we start getting an idea for shoot count and maybe some bud count early on and we start taking and making this funnel of reducing the uncertainty on what's happening. Then we move into the next part of the equation, which says, all right, now we're seeing flowering. There's certain events that happen around and around flowering, such as high wind events, various other heat events that can reduce that potential for fertilization, going from the flowers to the fruit. And then we say, all right, that's locked in, right? What's the next piece that's not locked in? Now we're talking about fruit set. And so we get to fruit set being able to say how many, what's kind of the, how good is our fruit set in terms of the compactness of the fruit? You know, what's it look like this year? And that gets locked in. Now what's still looking forward? Now we're looking out and saying, okay, now the berries are going to swell to a certain size, right? And that goes all the way to physiological maturity, which usually happens around somewhere like CAB, somewhere between 19 and 21 bricks. So that's still a ways out from harvest, right? And so that point, you get a max berry size. And so we have an idea then for what's the maximum potential berry size. And most of what drives weight loss after that is dehydration, you know, we call hang time, right? And that's what's happening from 21 on. So if you break the equation down into a lot of small parts, we can see when where we have uncertainty and where we don't. And when we can lock those in and when we can't. And then the cool part that I like to think a lot about is now, how do we think about the unknowns, right? And that's what you were asking about. And if we start thinking about the unknowns, you can look at the history of how a given block instead of vines performs in response to different climatic anomalies and whether anomalies that might occur and say, how much could that part of the equation move around? Like, you know, there are certain some places that are really stable out in that, like, for example, let's say sparkling, they don't even worry about the hang time issue, right? Because they're harvesting early in terms of bricks. That's an extreme example. But people harvest at different bricks, they're going to manage at different levels of stress. And depending on the block, these will make kind of like bring up and down the importance of each part of the equation in terms of both what could happen in the future and what has already happened. And so by tying all those pieces together, I think the thing we really like to say is you'll you'll forecasting as a puzzle, right? It's both an equation, but it's also a puzzle. And each piece of this, we have to understand all the different parts of the equation. It's not just the photos, right? We talk about taking photos and measuring, taking metrics from photos, but it's a lot more than that. That's a measurement, right? That's what we can see right now. But then we have to layer in a lot of the AI to detect the patterns of how weather and site your specific site interact with those patterns we're seeing on the ground from the photos to make the forecast out into the future. And so are you taking in like a weather forecast to to understand what youth experts think will happen? Yeah, so what we do is we take the history of any given site, looking at the best available data out there on 25, 30 year history at that site. The more historic yield data and performance data, whether that be all the way down to cluster counts and weights, we have the better we can pin down that historic data to what happened in the past. And so whenever we start looking at pinning those two together using kind of complex pattern matching, we find through the AI models, then we can start saying, okay, for your site, we're not saying, you know, you're down in car narrows, we're seeing wind is a big, we've seen wind events make a big difference in terms of fruit set. So that might be like, hey, on your site, there's a question beyond that, which is like, yes, that's a risk, but then, you know, from a risk management perspective, there's mitigation that may have been done also at that site, depending on the same with freeze and frost events early on in the season. You know, the same side by side, two blocks, one could have frost fans and one doesn't. And the effect both have the risk, right, the frost risk, but one may mitigate it differently. And so we also have to layer in the management that goes into this. So that's kind of the different pieces of the equation come from the climate and weather history. So I'm curious though, because like that's when Peter first asked the question, I thought it was more of like kind of like general kind of like temperature and historical things, but do you ever talk to you talking about very specific microclimate inside the nuance of a vineyard? So are you gathering temp and humidity as the tractor drives through or is it just a camera that's getting visualized? No, not right now, but that's not to say we, you know, we could use that data. You give us whatever we will take whatever specific of site data you have and use that. So if you have, if you have site specific weather sensors on your site, we use that. It's that we've set up the API, the software such that you give us more granular data. We make more granular predictions, right? And so that's sort of the way we work right now, right? The generic one we can do for anyone out of the box without even having it is we work with some of the best AI weather forecasting and historic data folks out there to try to say like what's the state of the art for being able to give a lot long, give a site and say what's been your history? And there's that's been improving a lot as well on that front. That's something AI has made a big impact on its weather, both historic and forecasting. And this goes back to the intelligence piece that I mentioned. So we invest more onto algorithms in the models as opposed to the sensors that collect weather, right? Everybody's got one. And we're not in the business of saying, Hey, you should pick this weather station versus that one or this irrigation tool versus that one. They're under utilizing the data that they have because the models are not trained to correlate actions. And this is something I've done in my past that's one, many other companies, which was how do you basically bring data that is siloed and glue it together and create predictive models. And big part of our vision on farming is to move from reactive to predictive yield forecasting is the first part virus detection is the second part, which is how do we take early symptoms of something and create it early warning system? If you use ways, ways is using AI in the same way that if there's a car crash, your forecast might be wrong. If you get into the car and there's a car crash that shows up. But it sets you expectations and lets you know where your risks are. And I think this is the problem with just taking a number. Estimation is like counting how many tennis balls you have in a bucket. That's just a guess. And some of us are really good at guessing and some of us are not really good at guessing. No business I know of can live off of guests and guest them. So it's underlying all the unknowns that are measurables using AI to fill into blanks on the unknowns and give us multiple scenarios that we can evaluate and then give us a range. And again, just like the business dashboards I've seen before, we give businesses three forecasts with a range of worst case scenario, best case scenario, likely scenario, your historical yields. And we think CFOs love this because this brings at least arranges so that they can make the right business decisions on. Mason, you mentioned sparkling to cab that's hanging for a long time. Does the customer have to input that in to know what the yield estimate is? Because if they're pulling it for sparkling versus pulling it for cab, it's going to be 20% different or something. So all this data, we have all the site-specific data. We know we work with customers. We know, hey, what is your coat? What rootstock? What clone are you growing? What trellising and training systems are you using? What is your specific site condition? How old is the vineyard? How are you managing it in other ways? So all of that is is useful information. And then the other piece is we try to prompt people, growers, and vineyard early on to understand what are the targets that are being set and what are the management, typical management approaches or mitigation things you typically do. So that's set up front. So we can kind of know which pieces may need to be elevated of the equation and which might need to be kind of down-regulated. So it's set up front what the use of those grapes are going to be? Yeah, that's right. Yeah, no, it is very specific. We are, this is, I think, what makes us pretty unique is that we are very, very focused. We have some of the best viticulturalists working in our company, right? Who have managed some of the best vineyards? And I think that is like We try to take it very seriously the one thing we want to avoid is partially solving the grape industry's problem and jumping around to other industries and partially solving their problems to to and so that's sort of we want to go deep right just as people do when they grow and manage grapes. So do you guys have any good case studies now that you've been around for a couple years in terms of the benefits so like a specific winery did something and they saw x amount of labor savings or whatever you know. Yeah we have a number of those and let me give it share a few of them so one of them is bench vineyards Allison Stelster and her dad are legends in Appavali they go back 40 years she was early a doctor of scout brought us in last year Matt Hardin and his crew manage bench they're great to work with everybody had the mindset of let's jump in and see what the data tells us as opposed to just basically again having pre notions on the very first scan of scout we discovered an acre of unplanned vines out of 24 acres so that's $400,000 an acre of land that needs to be farmed and it wasn't because anybody was bad it's just to go and manually walk that would cost you about three to $4,000 just to basically send a crew out there to figure out where everything is and what's missing so first scan of scout literally created a nursery order that literally out of scout they ordered the root stock and the right plants and those vines are now planted a year later and are ready to go you mean this wasn't an acre total like there are missing ones or just one continuous piece it was onesy two z's and that's what we don't get when you're walking a vineyard you're not stopping and saying hang on a second why is there a blank spot there are blank spot there are blank spot there everybody was amazed when they saw it so missing vines that added up to an acre out of 24 got it I was like it'd be pretty weird for them to just miss a whole acre block no no no no but but Peter let me let me be look your listeners need to hear this 100% of the customers we've gone to had the wrong data on their maps 100% okay so one of them which I won't name but he's a great white maker he's like good news bad news I fixed a problem with low yields out of this block good news as I solved it with scout bad news is the acreage was wrong I had the right tonnage I had the wrong acreage and I wasted a few years trying to fix it so if you're not operating off of data that is accurate you can have the big and best experts in the world they're going to make the wrong decisions we think the budgets are not correct if you don't have the right vine counts if you don't know how many producing versus non producing bonds you have your forecast is going to be wrong so we have foundational data that was incorrect that AI is better I promise you if you go to a checkout counter at any retail shop they don't have a cashier type in the serial number anymore right it's a barcode scanner because the barcode scanner does a better job faster job makes less error and we're bringing that to vineyards so that we can very quickly get how many vines do I have you know are they productive are they non-productive I'm confident in the next five years we will not do manual vine counts if you're serious about wine as an industry and a profitable business think about the asset class that doesn't get monitored so we get above that what we found is about four hundred to twelve hundred dollars in savings per acre again myelage big mileage vary and that's on cost of inputs it's productivity where people are not walking down the wrong row daily looking for things because our mobile app allows you to literally use GPS like ways and know exactly what vine you have to go to if you're kind of pulling it out or if it's your scouting our AI solution now has uses AI to tell you direct you to where to sample because what we found is people were sampling wrong some cases randomly in some cases their favorite vines and those vines may not be representative of the vineyard so now we're using statistics and it's all about precision guided activities as opposed to one to many activities that the the whole wine industry's been on input savings is another one we wanted a coolest thing and I love Mason to talk about this that we came up with is we can go into any vineyard and with less than three minutes identified the lowest 20% performing vines and then create a farm plan that fixes those Mason do you want to talk a little bit about that and the data I think it's a cool yeah I think it's a really powerful example so you know we have the ability once we start this is one of the things that I think some people asked us before is to say hey I like what you guys are doing but I can't manage every single plant like even if you know what how would I do this you're measuring at the plant level and we say yeah we're measuring at the plant level but when you measure at the plant level it allows you to accurately manage at any zone level size you want all the way up to the block to the portfolio of vineyards and ranches that you have right and so this is an example where we start with the plant measurements and we have looking at both the canopy vigor we do something called canopy volume index which we've we've created which I think is a really kind of even the great version of basically very index better than in dvi these sorts of things a lot that are out there and we look at the fruit so canopy fruit and we can kind of cut across those two pieces to say hey where do we see we kind of create a grid and say where do we see low performance across both of those where do we see high performance across both those axes canopy and fruit and we can start creating looking for zones of clusters of plants that fall into those different buckets and so once you start clustering into a low canopy low fruit let's say performance category and then you say I want I don't want to deal with the one or two that are hanging out by themselves because that's hard to manage and go to every single plant make sure I have at least 20 plants in a zone so I can just target 20 you know I'm going to go there and now there's return on my investment for going to that location we create a performance zone and you can decide how big and how dense how big is your zone going to be and how densely populated with that performance metric does it need to be defined by and so once you have that you can go in and we add an example once you have the plant level like I said measurement at the plant level you can go in and we had someone say hey I'm thinking of we found a low performance zone they said now what do we do well I maybe I need to pull it out and we said well we can tell you what the yield on a we can create any combination of plants we want to tell you the yield out about every one of those plants so we took that performance zone tell you what the yield is relative to the rest of the vineyard and we tell you you're losing 35% you have 35% less yield coming out of that particular zone take that and look forward four years and say how much is this going to cost you over the next four years and you can then justify whether or not a replant is going to make sense financially and so it was it did and that's what they did and I think it suddenly allows you to do a make a lot of financial decisions that are going to lead to better outcomes you know looking forward as opposed to just saying I'm going to wait and man let the whole block go down or I'm just going to let that corner hang out and be a weak performer for the rest of the vineyards life I think it really gives you this ability to scale in and out all the way from the plant up to the block level and manage that way as well it's interesting about all those savings examples and and quite the range of it I guess the for the trade audience that is listening to this I think the question's going to be like what is the solution cost and can you break it down into up front versus ongoing cost that you think it would would have yeah absolutely so again I'll start by saying our our business model is a subscription model roughly the cost are about $150 to $180 per acre we can actually price-provined but most people don't know how many vines they have but our cases where they have sparse wines rose that are much larger and we'll we'll accommodate those their volume discounts above 50 acres that significantly bring that price even down and also we have neighborhood discounts things like virus make sense to do with your neighbors and we have avy-wide discounts that we've rolled out and that's been very successful in apple and Sonoma also what's important to talk about is that the upfront cost is standard off-to-shelf smartphones and cameras that that cost under $3,000 if you're doing one scan will even loan you to equipment so you can kind of get a get a get a feel for it without any upfront capital cost we do this for vinyar management companies and many of them are already equipped they have the equipment they can come in do the scan for you if you wanted to come in and use Scout completely purely as a service okay so and then in terms of benefits like how quickly and what kind of ROI is there your customer seeing our eyes within the first year and again I need to go back when you're charging 150 to 180 per scan we offer two packages one is a two-scan package and typically this is inventory and virus virus is important to do late season post harvest so you catch all the red leaf and be able to use it included in those scans is a mobile app that allows you to go and make edits and has a lot of the functionality that our GIS sent in all those stuff a product have so that you don't need a separate app that does vine by vine manual entry we combine the AI scanning with the manual entry we also have a pro package that is for a state that are really driving performance and those have six scans or more the timing of those scans is important if you're again really obsessed with quality we can do pre-imposed pruning scans we can look at your crop evaluation you know in terms of vigor and fruit at various times and also do the virus scanning so it's almost a season in pass. So the starter and the pro are the ways that we go and the only difference in price is on the number of scans. In typical customer is using how many stands? I think what they do is they start out with two. We're running a promotion right now that if you're a monocustomer we give you an extra scan so that you can get comfortable so you can maximize that and then the graduate to six because they see the value of the product and again for a state wineries, a state programs makes no sense to not use this product again. We believe in the ROI that it provides our minds as every scan should pay for itself and then some. Two saying six per year. So if it's a hundred fifteen acre then it's what nine hundred bucks a year. Yeah and that's not saying again if you're saving twelve hundred dollars an acre, Peter and that's verifiable because again you got to not think about the people are spending this money today. What we found unfortunately is they're spending it in labor and they're spending it with VMCs and they're spending it in quality hits. We've shown that if we can improve a block that goes into a premium skew and not have that vine be declassified what is the business savings? So a lot of the budgets that we've set up let's be honest and vineyard management is labor centric is not preparing us for the future where we have to invest more in technology to offload monotonous has to nobody likes like vine counts so that that's not part of everybody's budget. You said an acronym just now VMC is that vineyard management. Just in case people didn't catch that. That's a huge one for us. We have some great customers there enterprise vineyards in Sonoma, Phil Caterian and team Matt Hardin and Napa that I mentioned. Pinya vineyard management number of other vineyard management said look at this software has been essential because it helps them scale quality across multiple customers. It allows them to be smarter around what their people do versus you know and keeping track of all these different sites is is a nightmare. It literally is a nightmare keeping track of just all the all with these outdated maps and the wrong vine counts that we talked about. So do they then reduce their price to their customers if they're using less labor? So I don't we don't get involved in that that that's their prerogative. I think what it is is they're delivering a better service in a smarter way and what we've seen is the quality of service for the customer goes up because they're they're basically instead of more passes human passes. The human passes are much more efficient. You collect ten times as much data. Guess what you're spending more time analyzing the data and acting on it than just spending so many calories just collecting the data and not being able to act on it. We've heard from people over and over again. I get the NVDI scan. I look at it. It goes on to desk, but I just don't have enough time to go act on it because it's not actionable. It doesn't tell me what I need to do. I want to kind of highlight a new product that we announced yesterday. It will be in beta in July. It's a Chad G PT and Gem plugin called Scouts for Vineyards. This helps the reference side of things. We give the accurate data, but we don't want people to run back to the office to look up their viticulture guidebook on what this diseases, what is Eska's signs. So now all of that is right there in your mobile phone and you can use a prompt to ask more questions about clone and root stocks and weather information and GDD days. So I want to make sure that your listeners understand where AI company, we're not just a picture and camera company. We're about how do we bring productivity to people who walk to Vineyard. That's what we all need to talk about because labor costs are not going to stay where they are today. The number of people who are in wine are not going to increase. We have a massive labor problem that's coming and Mason sees this on the front line as he teaches in the university as well. So how do customers hear about you guys? So word of mouth. Number one thing for us is word of mouth. We've been kind of low profile in the last two years intentionally and it's analogous to Tesla's self-driving. We have to recruit the rights that are customers to train our models and make sure that we're working with the absolute best farmers. And then now we've expanded to a public generally available solution. People recommend us. And then also we get that feedback through shows, trade shows and industry groups. So we were at the Sonoma, Sonoma Vitech conference recently. We're at the growth full ahead of the curve with the Napa Vine growers. We did a great pot or webinar with the win network. Which of those have been most effective in terms of converting to customers? It really has been the industry specific trade shows and then also kind of the webinars. Also the partnership with Monarch has been very helpful. It's led to a lot of both joint customers who want to get ahead and use technology, their technology enthusiast, if you will, that they believe the same way as we do that the industry needs to evolve and change. So I would say those have been the best lead gen sources and we have a pretty cool website that is very transparent. Our pricing is on the website. All the product descriptions are our website. We just believe in transparency and being open and educating the industry. So the wide industry is notoriously antiquated and you guys are very revolutionary in terms of pushing on AI and the vineyards. So what are the barriers that you hear from your customers for adopting your solution or potential customers, I guess I should say? So I think look, that's a big generalization. I swear to God and again, I really cannot. I've shipped a lot of products. I've worked with a lot of products in different industries. I've never had a product that is so well received when people see the demo because we've literally had windmakers tell us 10 years ago I asked Google to build this when I saw Google Maps and saw Google view. We've had VMCs that literally did a double take and brought their owners into meetings. I think skepticism and habit happens in growers. Gores have the right to be cautious. And this is why we encourage people to do a small paid pilot on handful of acres, you know, 10, 25 acres. Get familiar with the product before they expand and land and expand has been great for us. We land within the customer, prove the technology out, have it paid for itself. And then we have a discussion to see how can this be adopted in a broader scale within that organization. So there's no noticeable barriers for people to adopt a solution. Number one barrier is today's budgets are built for manual labor, the old way. They look over, look at the budget and say, I have it in the budget. Can I can I take this investment and disinvest and put it in and make a case for it? And we've seen that. I mean, we're displacing a lot of point products that people bought that are partial solutions. We're also showing the results on how much fat again, four times faster, 10 times more data, like four times faster. Humans can run at 10 miles an hour and collect data. I don't think any technology does that. And it's only going to get better. We're going to get faster, collect more data with every revision. So it's face to ROI discussion as opposed to the cost discussion. And so you mentioned the chat GPT in July. What else is on your product roadmap? Mason, why should you take that? Yeah, sure. No, I think one of the ones we've been putting a lot of effort into, which we have coming at its invader right now that's coming up is both the sampling side. And we call it this is AI guided sampling. And that helps us a lot to say like he is example of virus our goal. What would be perfect? And I say this, you know, would be is if when you did sample, you had a hundred percent hit rate on virus. Now we don't want to see any virus anywhere, obviously. But right now what happens is you get a lot of samples sent out that are that are negatives. And oftentimes, you know, you spend a lot of budget doing that. If we were doing our job right where we were making recommendations, we tell you where to sample and we're very effective at it. So that's that's the one thing is the AI assisted ray eye guided sampling and other pieces this ability to do zoneal type of management. And so that's one we're working on and have it in beta where that's another release coming up soon is the ability to take any level of zones and what we call tags. And so tags are the the ability to group plants in any way that makes sense to you. So for example, in the virus case, you might have had some positive hits on red blotch last year, right? You did some rogging and then that would become a red blotch positive cohort, which is all take the same tag for that year. And what we can then do is we can start saying, hey, let's look around all those plants and see if we see any impacts on the canopy or fruit performance in neighbors. Right? So this helps us guide that sampling that I was talking about earlier to say, where do we want to send you to get the most likely bang best bang for your buck in terms of sampling? So that's I mean, we're sticking on those three. We still see a lot of room in terms of inventory, plant and fruit performance and yield and virus. But there's a lot of what we're thinking about now is how we build workflows that go into and so you can say like, hey, we found this pattern. Click a few buttons. Let's send out some workflows and work orders to people to go sample specific plants. I pull out my mobile phone. I go into the field, know exactly which plants to go to go there, type in whatever, you know, the data and observations I saw, take my sample, tag it, send it back. It works its way into the system and create kind of a closed loop where we can really like see it into to end from the images all the way through the management and the action. - Mason, quick follow up. So you've mentioned a lot of times you send out samples and they're negative and it's a waste sampling. Is there a KPI? Is there like a percentage of hit ratio? Like what is that currently in the industry and what will your solution bring it up to in terms of effectively detecting samples? - I don't know if there's the industry wide benchmark. I wish there was. What we're finding is that most people doing event have adequate budget to test their symptomatic finds. So what they tend to do is they use it by proxy which could be very misleading. Like take five leaves from five plants and send an example. And at that point, you're actually the results won't be accurate anyway because it could have been one or five. So I think this is an area that we're establishing something called the health score. Health score is everybody, every block starts with 100 and if you have confirmed diseases, it goes down. And then we give you a dashboard on which blocks you need to focus on. We have symptomatic or observations that you make versus confirmations and scout tracks the entire life cycle from sample taken to a PCR test or a lamp test that was run. We create a serial number for every plant and every space and we use that as the primary key to keep track of where the disease was so that we can correlate it back to a zone or as Mason said, the tax. So part of the value is that we have a 7,000 strong photo database of confirmed disease lab tested photos. That number keeps rolling. These are photos that are not just random photos. There were photos that were back tested to actually be negative or positive. And so our goal is to get to 60, 70, 80% in the field confidence level on a per sample basis. So that right in the field before you leave it and you can see what's the confidence that this is going to be leaf roll three or red blotch. And then I can decide whether I want to spend $40 to get that 100% or I don't. Does that make sense? - Yeah, it makes sense. That was a good answer. I mean, I just sounded like a great example of a potential case study. So obviously you guys have showed a ton of information about scout and what you're doing in the vineyard as we go to wrap up this episode. We'd like to end on a personal note. And we are curious for each of you. What is your most cherished bottle of wine in your cellar and more importantly, when do you plan on drinking it? - Absolutely. So let me go first. I have a 2008 crude champagne, which I'm saving for the first time we get to 10 million vines. And hopefully that will be soon probably ended this year next year. - Awesome. - We'll be right back. - Oh man. We have a lot of good customers and we've gotten a few very special bottles of wine. Typically wait till the end of our software development. The busy time for us is the same time that's busy for the people that are growing grapes. That's when we're working the hardest too. So we wait until harvest. And then I think I'm gonna be popping one of those. It's very special to us to see the bottle that has been at the same vineyard that we will hopefully make the future ventages at. - So our harvest parties every year for the engineering team is the bottles of our customers. And then hopefully in the next couple of years, especially with the red wines, we'll be drinking actually and looking at the maps and the metrics. - Awesome. Well, thank you both for sharing. We really appreciate it. And helping educator listeners on what scout can do for them and the vineyard. - Yeah, thanks for having us. - Thank you so much for having us. - Hey listeners, if you love the show, support it by buying a show notes book. They not only compile two years of episodes, but also organizes them into themes for better learning. They can be an inspiration to listen to or relist into an episode or provide a quick reference of the key learnings from a show. Go to xchatto.com and click on the store page for easy links to buy. Thanks for listening. [BLANK_AUDIO]