How AI for Agriculture Can Outsmart Volatility

Here’s the paradox we’re living with: food prices keep climbing for American families, yet American farmers are facing shrinking profit margins. As our guests in this episode point out, the volatility of agriculture and the inefficiencies baked into the supply chain often leave both sides as price takers.

But there’s a tool with the potential to flip that script. AI. It’s everywhere in the headlines, but we don’t always talk about the real, on-the-ground impact it can have today.

In this episode of the S2G Podcast, Chuck Templeton sits down with Shail Khiyara, CEO of SWARM, and Adam Greenberg, CEO of IUNU, two S2G portfolio leaders who are utilizing AI in very different ways to help the food system navigate volatility and make smarter, faster decisions. They dig into what it takes to drive adoption in a sector famously wary of new tech, what the farm and food system of the future could look like, and the partnership models that could change the game.

Chuck: Most of what’s in the news these days are dominant AI stories about how Nvidia, OpenAI, Oracle, or Google are doing hundred billion dollar deals with each other. And I have two of our entrepreneurs here, Shail Khiyara and Adam Greenberg, whom I’m excited to talk to a little bit about how all this is impacting food and agriculture, because we believe there are tremendous opportunities. 

The agriculture industry has low margins. It’s heavily regulated, with long sales cycles, perishable products, an oligopolistic market structure, and built on physical assets and annual growing cycles.

And so there are a lot of challenges that sort of move AI out from quote unquote just a technology, to really thinking about how it is applied to industry. And food and ag has had some challenges lately — depressed commodity prices, persistently elevated input costs, higher interest rates, and climate change. 

And so with that, I’d love for each of you to quickly introduce yourself and talk a little bit about how you are applying AI. And Shail, maybe let’s start with you. 

Shail: Yeah. Happy to do that. Thank you for the opportunity, Chuck and the S2G team. So look, I actually started out as a civil structural engineer, and I’ve always had the instinct to build things that last. Early in my career, that meant bridges and buildings; later, it became systems and companies.

Over the years, I’ve worked with public and private companies across intelligent automation, AI, and transformation, helping organizations figure out how to use technology, not just to do things, but to do things in a different way. And what pulled me to SWARM was this idea that agriculture is the ultimate complexity puzzle.

You’ve got biology, you’ve got logistics, weather markets all moving at once. And it’s one of the few industries where volatility isn’t a bug, it’s the actual environment itself. Every day, farmers and Agri-Food companies face hundreds of interconnected decisions. What to plant, where to process, when to ship, how to price, how to adapt when something breaks. And most of these systems are not designed for that level of complexity.

SWARM is about helping farmers and Agri-Food companies reason through that volatility.

SWARM is a decision intelligence company, and we use AI optimization, machine learning, operational research, and what we call agentic reasoning to help teams navigate that motion. So instead of static plans, SWARM creates living models.

Think about digital twins that can learn, that can stimulate, and recommend the next best action as conditions change. You think about it as an intelligence layer that connects everything. Data from your fields, your plants, your logistics, your markets, all reasoning together. Decisions happen faster and with more confidence. So at SWARM, we feel volatility isn’t the enemy, it’s actually the signal. And that’s what we help customers with.

Organizations like Ardent Mills, the largest baking flour manufacturer in the US, AgroVision, the largest berry producer in Peru, Wilbur Ellis, and various other companies use SWARM to optimize supply chain logistics, workforce planning, harvest allocation, and drive operational efficiencies in their organizations.

Chuck: Adam, tell us a little bit about your background, pun intended, deep roots in the greenhouse industry.

Adam: Yeah, appreciate the question, Chuck, and thanks for having me. My father’s been in the plant business my whole life. I used to compete for attention between the orchids and myself while going to school. For me, it was always personal when it came to plants.

I worked at Amazon back in the early days, 2010, 2011, and learned a lot about what operational excellence looks like there. Started a water business and reverse osmosis and membranes, and then started this business in ways to optimize both the value chain, the supply chain, and the greenhouse itself.

When we got started, everything was manual and disconnected. And so IUNU (spelled I‑U-N‑U) uses machine vision and off-the-shelf cameras as well as AI and integrated data to standardize, analyze, optimize, and predict for the farms across the whole greenhouse industry. So whether you grow tomatoes, cucumbers, peppers, lettuce, or whether you grow poinsettias or orchids, we work with growers to then use their information and their current data on their farm and integrate it with new types of data from machine vision and AI to be able to deliver better forecasts, which allow them to do things like supply and demand matching. We reduce volatility on the farm as well when it comes to predictions.

So what does that mean for them? That means they have a lot more certainty when it comes to the outcomes, and their sales teams can then use that yield assurance, if you will. So we sell that certainty as a yield assurance, and that cuts their variance in forecasting, which allows them to get better pricing.

For every week that you can predict in the future, you can get better prices because you can lock in those prices. The second thing that we focus on with them is at the farm gate. How do we optimize that yield? So how do you measure plants every single day? Currently, without us, they use a tape measure, a caliper and a string, and then they enter everything manually, and then they look to graph it over time. 

We automate everything they do manually, and instead of doing it 10 or 30 plants out of 200,000, we’re helping them do it 10,000 or 20,000 a day. And that statistical significance and sampling allows them to be far more accurate in both how they steer and grow their crops, how they manage their labor, and also how they predict the future with their sales team.

And that sort of locks it together into that operating system for a greenhouse that allows them to reduce risk, but also optimize outcomes.

Chuck: It’s interesting because digitization for ag took a while. It certainly was a slower adopter. And ag does seem to be adopting AI a little bit slower than some of the other industries out there, although there are signs that it is accelerating. We see some of the challenges in the past, the price point of some of these tools, and trying to figure out which ones actually work and have real value. You have demographics where I think many people don’t want to learn new things or are comfortable in their ways. They’ve worked for a long time, a lot of siloed point solutions. And then, just deterministic software versus AI and its predictive software. Shail, you’ve talked in the past about new ways of thinking and addressing some of these complexities. What mindset shifts are required to be able to do things like digital twins as a concept that everyone up and down the organization can understand?

Shail: Yeah, look, that’s a fair question, Chuck. I think adoption in agriculture has been slower. But I like to think it’s not because the industry lacks ambition. I think it’s because it’s really built on reality.

I often say that agriculture is not really slow to adopt. It’s just allergic to nonsense. And really getting down to the heart of the matter and showing the value to the customers is critically important. So digital twins and AI bring a completely different mindset.

For years, we’ve tried to manage the uncertainty in agriculture with static tools and fixed plans. We have managed it with spreadsheets, forecasts, and, I would say, a lot of hope. And resilience in agriculture is about decision options, not hope on a spreadsheet.

And when you take something simple, a grower trying to decide when to irrigate, or a co-op trying to decide which plant to run next week, or a harvest, trying to find the best possible allocation to maximize profit and minimize waste. With digital twins and AI, you can see the impact before you act.

You can test a dozen what-if scenarios, weather changes, labor shortages, and know which path gives you the best outcome before you spend a dollar or lose a day. And I think that’s critically important. It’s not about replacing experience; it’s about amplifying it and pairing judgment with speed, intuition, and computation.

And that’s when AI stops being a tool and starts becoming a partner. I think the real edge in agriculture won’t belong to the biggest players anymore. It’ll really belong to the fastest learner, the one who can turn uncertainty into advantage. And that’s where digital twins and AI play a huge role.

Chuck: Yeah, I think there are tremendous opportunities with the flexibility of what sort of AI brings to these operators at the front line. Like you said, I don’t think it’s about replacing people. It’s about augmenting and enhancing their capabilities so that they can make decisions that are accurate. There’s a tremendous amount of efficiency gains in that.

Adam, from your vantage point, working across hundreds of facilities out there globally, what have you learned about implementation, and are you seeing growers adopt at a much faster rate and seeing the pickup here?

Adam: Yeah, it’s a good question because the adoption rate is relative to the products around the market, and the level of, let’s just call it uncertainty around who to believe is becoming harder and harder. So you see two different aspects of this. One, farmers take the largest risk of anyone every single year because they literally bet the farm every year on these crops, right?

And when they do that, we’re seeing that people are intentionally spending time to trial and adopt technologies that include AI because they know they need it, but they don’t know how they need it. Because there are so many different companies now using the word AI, it’s almost like what ​“cloud” used to be.

The word AI is becoming so buzzwordy that it creates a lot of noise. And so what we’ve really found beneficial is we don’t focus on AI anymore. We focus on the outcome. AI is good for a lot of things. The way we talk about it is based on yield assurance. If you have X amount of volatility, we want to cut your variance in half.

And we can stand behind it financially. We can stand behind it through our product. And by having people feel like you understand them, and you’re just using AI as a tool to solve some of their problems. That’s how we spend most of our time communicating these days.

Because at the beginning of the AI conversation, it was easy to talk about AI and have them get excited. Today, AI has been used so much that it’s just another part of the word salad. And so we’ve had to focus on AI as a tool and help them get to the outcome they want and work backwards.

Chuck: Yeah, I feel like internally I’ve tried to use AI as an expression less and really focus on what the outcomes are, what the changes that are happening are, so that people can really think about it from the work context that they’re in.

I think it’s that whole piece of here’s this general intelligence tool that can do everything to being deliberate about going after those specific needs and challenges.

Shail, how do you think companies should be starting to embrace AI? 

Shail: Yeah, a lot of it starts with really addressing the individual processes that they have in the organization. Of course, one would say a lot of it starts with data itself as well. In some instances, agriculture data is siloed.

So it’s about having access to that data. It’s about being able to normalize and rationalize that data. And then it’s about identifying what specific business problems you are trying to solve, whether it’s supply chain, whether it’s logistics, whether it’s workforce planning whether it’s harvest allocation, as an example.

We work with companies that have various maturity levels of problems that they have within the organization. And we go in and help them identify which particular ones they want to solve first. And often it’s either problems that are impacting margin, but the data is not in good shape, or there’s low-hanging fruit and the data’s in great shape, and you can apply simple AI capabilities to that. For example, a rag-based LLM that is very specific to your organization that draws all the information into a central repository, and you’re able to query it and get information at a much faster pace. We see varying levels of problem sets within organizations. I do think there is a strong interest in the market from organizations wanting someone to come in and help be a partner rather than a software vendor.

Don’t throw software over the wall, but partner with us to figure out which particular problem sets should we solve first.

Chuck: Yeah. And Adam, from a multimodal standpoint, you see a lot of different data types. You see a lot of different silos of data throughout an organization. How do you help bring that together to create decisions at the frontline? Because that’s one of the real advantages to AI these days, it does allow you to look at, again, very diverse sets of data and make logic out of them.

Adam: =I think it both presents a problem and an opportunity. The major problem, being so disparate, is the integration problem and the standardization problem, where people from the input perspective really struggled to put it all together. Initially, we struggled to find ways to standardize it and put it together because people would speak different languages and have different metrics, but they were all trying to say the same thing.

And they’re all trying to measure the same thing, but they all do it in different ways. And then on the output side, we find that the outputs originally were moving from a 2010 BI dashboard style world to a, I want immediate gratification, right? And so people have higher expectations for the products that they’re getting.

And so not only do you have to integrate it from a, let’s just call it multi-experience perspective, they want to be able to communicate whether it’s verbally, whether it’s written, whether it’s on their phone, or whether it’s on their computer. They want the fastest way to get the insights.

And they no longer expect to do the analysis. They expect you to do the analysis to make it so that they can just get a quick answer to be able to unlock time. And in business, for business enterprises, you really have three options. Two, specifically, you can save them money, or you can save them time.

And then on the third option, every once in a while, you can say peace of mind, but that’s hard to sell on. And so for us, it’s about how we are going to save them time and how we are going to help them remain profitable and be successful for generations to come? And that comes from an experience now with a much higher expectation.

As much as everyone says, ​“Wow, it must be easier with AI”, the expectations from the output perspective have gotten a lot higher, and you no longer have the ability to ship it quickly and hope and break a bunch of things and hope that it works. You now have to get it right the first time because the expectations are so much higher than they’ve ever been before.

Chuck: Yeah. And then how do you standardize that sometimes across multiple different customers or customer types, that allows you to scale as a business as well? Because it is so adaptable and so quick-moving that, by the time you’ve locked down the feature set, you’ve rolled it out, 3, 4, 6 weeks later, there are new additions you can do to it. And figuring out how to manage that rate of diffusion out in the market is just really tough. And one of the things I want to talk about is the concept of this idea, and this isn’t my language, but I’m embracing it- there’s AI-resistant, AI-neutral, AI-forward, AI-first, and AI-native, right? That’s the spectrum. What’s giving companies confidence to move from neutral to forward to AI-first? Adam, you were doing machine learning out of the gate and have transitioned more aggressively towards other forms of AI. And then your customers have also been on this journey. What has it been like for you guys to go through that journey?

Adam: For us, it’s been what I like to call the breadcrumb trail, which is to get to the outcome, you have to help people get there, and it has to be their excitement and their idea. They don’t always believe what’s in the imagery. The analysis says it’s this. I think it’s that reconciliation happens through the breadcrumb trail of, okay, let’s go measure it together. And so you help them go do it. Or in our case, let’s surface the imagery. They can count the number of tomatoes. They can count the number of, oh, you don’t have any ripe tomatoes to pick today.

Here’s some examples of your imagery of where you have 5% of the plant ready to harvest. Then once you do that, two, maybe three times, they stop doing that, and they move from being AI-reluctant to AI-excited. And then once they start to get excited about it, if you can win by giving them actual good answers two or three times after that, then they almost become AI native thinkers, right? So you have to breadcrumb them through it. 

I say to a lot of people now, we’re no longer selling software as a service. We’re selling software and service, right?

And in farming, it’s all about the relationship. And so we have to be focused on the service to be able to give them the outcome they want to help them become AI-native. If you say, what do you want to be? Out of all the ones you just mentioned, do you want to be AI native and be able to get the most out of AI? Everybody will say yes, but if you look at the theory of diffusion, or Crossing the Chasm, you’ll notice that only 17% of the population are innovators, early adopters. So how do you get people there? You have to use the breadcrumb trail, customer success relationships, subject matter experts, and being right a lot with AI to be able to move them from reluctant or late adopters into power users and AI-native. And that’s the responsibility of companies like ours to help get them there.

Chuck: Shail, you guys started off a little bit more AI-first, if not AI-native. Talk to us about that. Because you’re talking to companies that are maybe AI-neutral or maybe a little AI-forward, and you’re trying to convince them, what’s possible. But how have you gone to market and really had to interface with these large organizations to help them think about moving towards AI-first or AI-native?

Shail: I would echo a few things that Adam just said. I think the vertical-specific AI is the single biggest differentiator for us in the market. Being able to understand the space, being able to understand whether it’s a fruit harvest scenario and we’re moving tons of fruit, or whether we’re moving two and a half million chickens a week, or whether we’re moving 15,000 people a day to go pack, pick, and ship berries.

That SME expertise of vertical-specific expertise is critically important. But when I talk to customers, there’s a trend that I’m seeing of how customers are moving to becoming more AI-native, and this trend really falls into sort of four buckets, which I pulled together as an acronym called mate. 

The M stands for multimodal. Customers want to see that AI just doesn’t read data. It can actually see and sense data. So it’s about drone imagery, it’s about soil sensors, it’s about satellite weather.

A is clearly agentic. This is where AI starts reasoning, and it doesn’t wait for you to ask. It runs scenarios on its own. For example, how to adjust a feed when commodity prices spike or rerouting trucks when a storm is about to hit.

It’s not passive intelligence, it’s active partnerships. So that falls under the A bucket. The agentic. 

T is for trustworthy. This is coming up a lot with customers. They just don’t want a number. They just don’t want the optimization. They also want to know why, the explainability, the context around it.

And this trustworthy AI explains the reasoning, shows the trade-offs, and learns from the operator’s input. And then the E is really about interoperability. No one wants another silo. They want it edge-enabled as well. We’re seeing intelligence move from the cloud to the combine itself. Literally, and that means the decision happens in real time. Moisture alerts, for example, before spoilage. Routing changes, before delays. So that’s what I’m seeing, is that acronym of MATE in the market.

Chuck: What are the keys to really accelerating the adoption of these systems? Adam, you talked a little bit about breadcrumbs, but elaborate more on what’s going to be the breaking point to where suddenly you move from that AI-neutral to AI-forward to AI-first? 

Adam: You are either going to break your own system and processes because somebody’s going to retire, or you’re going to try to scale, or you’re going to prepare yourself for the ability to scale or the ability to handle your subject matter expert to retire. The average age of the growers above 55 in the greenhouse industry it’s even higher.

These are the experts. When they retire, your kids aren’t going to school to become greenhouse growers who want to work seven days a week, 360 days a year. So you have two options. You can do it through pain, or you can do it through excitement, innovation, and opportunity. Both ways, you end up with the same outcome. And so the way we talk about it, from just getting them there, is if we can solve some pain points for you that are quick and easy for you to get excited about, what other problems would you want to solve?

They build their own playbook. And when you peel back every layer of the onion to them they then lean in more and more as it keeps doing what it needs to do. And then I think from a human in the loop perspective, it’s important.

Because if you get one or two things wrong from an AI perspective, they use it as vindication, and it loses a lot of trust. And so when we deliver a lot of our LLM wrap solutions, because they hallucinate, to get that right has been a lot harder because it has so many ways it can go wrong. And so you have to be able to initially work with them to get good answers and guarantee good answers before you get to something that can give wrong answers one out of every 100 times. And so it’s that building of trust. Because people have been trying to sell them snake oil for 4,000 years. They’re very skeptical inherently, and they should be because of how much at risk they are every single year when they do their crop.

In order to build that trust and maintain that trust requires really a multimodal approach of helping them both with imagery to gain trust, being on site to help them, and also working with them, with their data to be able to see the power of it. And that’s oftentimes counterintuitive because people think that technology should mean that it’s easier to be able to deliver.

But the reality is AI is a tool. It’s still the relationship and the people in agriculture that make the decisions.

Chuck: We recently had an event here in San Francisco where Waymo hit a cat, and certainly a terrible thing. But at the same time, you look at the stats on these autonomous vehicles there are far less accidents than when humans are driving them. So there are places where these machines and these systems work, but the perception is that they don’t, because of these highlighted issues that come along. Shail, how are you seeing that?

Shail: I think it’s a little more than human-in-the-loop. I think when we look at breakthroughs that are going to accelerate adoption, I agree with the human-in-the-loop aspect of it. And look, I’m a big believer that silicon needs carbon, right? AI needs humans. Humans are not going away. But I’m going to go out on a limb and say that the real breakthrough isn’t another algorithm or app.

It’s happening underneath that. And what I mean by that is we’re starting to see operational research, machine learning, predictive analytics, and neural networks start to weave themselves into a physical backbone of agriculture. The mills, the silos, the plants, the ports. And that convergence is creating something new.

I like to call it the cognitive grid. Not too different from a Waymo example that you shared, Chuck, because there is an underlying grid aspect to Waymo as well. But this cognitive grid is a living network of intelligence where infrastructure itself, agriculture itself, just doesn’t plant, it starts to think. A mill that can learn from every batch, right?

A plant that optimizes itself mid-shift. I’m talking about a manufacturing plant or a port that anticipates congestion before it happens. And the future of agriculture isn’t just automation, it’s contextually aware action. And that’s critically important. Sensors, models, digital twins, we touched on all of those, will run quietly in the background, learning the rhythm of agriculture, the rhythm of land.

Before a farmer or Agri-Food company makes a move, the system will have already tested hundreds of scenarios, rain, price, yield, labor, and surface the next best step. It’s quiet. It’s quite profound, and I think it’s happening right now. And for the first time, the world’s food system isn’t just producing, it’s beginning to learn.

Chuck: How can AI change the economics of the small farmer, the grower, even the larger grower? Do you see AI being able to bring solutions that help farmers one, make more money , and two, ultimately control food prices.

Adam: Not only do we have to look at what’s going to change, but we have to look at what’s not going to change over the next 10 years. Consumers are always going to want better prices, and farmers are always going to want higher prices. So what that means is that you’re going to have to find efficiencies.

Our hypothesis here is that the value-added services of AI have to drive margin gains. And they have to fundamentally shift the economics for farmers to be able to maintain the profitability they want to target and where they want to focus. And so from that perspective, we expect it to shift the economics in a way that can not only optimize yield, but also lower costs.

So you have to find a way to make sure that your costs are well-handled and efficiency is maximized. And AI is that. In Pareto, humanity took the 80% of value with 20% of the effort because no person can measure every single data point.

Now with machine learning, AI and LLMs, you can get that final 20%. That took a lot more understanding, a lot more context, and started getting those efficiency gains out because that’s the only answer. To be able to maintain profitability while you have the consumer world hammering on your head, saying your prices are too high.

Meanwhile, farmers are saying, I’m just trying to eat. I can barely put food on the table. And so something has to give. And the thing that’s going to give there is the fact that we have a lot of waste in our value chain of a lot of waste in our supply chain, and that’s how we go about solving this problem.

Shail: I agree. The profit won’t come from scale alone. It’ll come from precision and having the ability to reduce waste, increase efficiency, drive resilience on the farm, and more importantly, participation, right? When intelligence becomes scalable, where small producers and farmers co-ops can have access to the same decision power that global players do with AI I think that creates a significant differentiating advantage for them and improves economics as well.

Chuck: Which ones does it benefit more: larger organizations or smaller organizations?

Adam:  I think that’s a hard question to answer because on different time horizons, they help them differently. So I think we have to decouple the question and add the x‑axis of time, which is the people with the most amount of data, and new types of data will be more successful upfront.

But as you have more data, that will unlock the ability to get more precise, and the smaller farms will benefit from it, and then their localization will benefit more once those models are perfected. So the more data you have, the smarter and faster you learn, the better off you’ll be in the short run, but the more you can then take that learning and be just as economically feasible, viable, and profitable at a small scale. And the reason for that is from learning. I don’t know the time horizon of that, but the inevitability is there.

Now, to go deeper into your question on supply chain, this also leads to a lot of disintermediation of the supply chain and value chain. If you are going to stay in business as a distributor, you’re going to have a value-added service. And that value-added service is going to have to be continuously more and more valued to be able to maintain either the stronghold or the relationship with the customer or with the solution provider.

And thus, the more we see the democratization of knowledge and agriculture and getting more precise, you’re also going to see that relationships from producers and solution providers will be directly with growers and farmers. And so the supply chain itself and the value chains are going to get a lot shorter. So who loses in this? It’s the people who are touching products for short periods of time, and who are in the middleman situation. The people who are winning are going to be people who are the actual producers and the consumers and maybe even retailers.

I think the inefficiencies are also going to get removed from this value chain and supply chain based on the fact that we’re losing probably 20 to 30% in our supply chain and value chain due to waste, that will also disappear due to the ability for platforms, technology, and AI to remove that waste.

Chuck: Shail, how do you see successful partnerships in this? Because I know you primarily, at least for now, focus on a lot of the larger organizations. What does a successful partnership with these large organizations look like?

Shail: They vary, and there are multiple flavors of that. I do see opportunities where equipment and AI alliances come together. Think about equipment manufacturers partnering with AI firms to drive models that understand field variability, machine behavior, and maintenance patterns, as an example.

We see opportunities where processors and optimization startups come together to focus more on production planning, routing, sustainability optimization, the one that I was talking about related to SWARM focused LLM alliances as an example. Growers, processors, logistic firms, research institutions. The University of Illinois is doing some interesting work. Virginia Tech is doing some interesting work as well, particularly around building large language models that speak agriculture and are trained in the ontology and lexicon of ag. And then there’s cross-value chain consortium opportunities as well, multi-stakeholder alliances.

We’re talking about inputs manufacturing, logistics, finance coming together to create this collective intelligence feeding into the broader cognitive grid that I was talking about. So multiple flavors of alliances out there.

Adam: Just one thing that I think would be really important for your listeners to hear. The food supply chain is fascinating because, yes, there were a lots of problems in it, but as we start to see the collective groups work together, for those who are listening and don’t know about this, the retailers are going out to producers —  like Pepsi’s going out to farmers, Walmart and Whole Foods are going to growers, right, in greenhouses, and they’re setting up 10-year long-term supply chain agreements where they’re offtake. So just as we saw in the power purchase agreement, in energy, we’re seeing produce purchase agreements that guarantee outcomes based on yields that you can deliver, and they guarantee pricing that then allows us to start doing really consumer-driven agriculture. We know consumers want more of X as a retailer, we’re going to give an offtake agreement for 10 years to this farmer. I think that to me is something that a lot of people don’t understand that’s already happening.

Think of it as demand-driven pull as opposed to a supply-driven push that is starting to shift inside of the produce market in the way that these growers now are able to grow with guarantees because they have these long-term offtake agreements with these retailers to be able to try to meet consumer demand. Because it’s far higher for fresh local produce than it is for supply.

Chuck: That’s our investment thesis, actually, is the consumer-led evolution here. So thanks for reinforcing that from our standpoint. What does success look like for each of you? Like when you’ve done what, how are you like putting a flag in it and saying that’s success.

Shail: Success for me is simple. It’s proof in the field. It’s a grower saving money because they adapt faster. It’s a mill hitting service levels because the truck stopped sitting idle. That’s success. When intelligence moves the needle in the real world, that is absolute success. I always tell my team if it doesn’t make the decision clearer or faster or improves, what happens next after that decision? It’s not intelligence, it’s noise.

Chuck: Adam, what does success look like for you?

Adam: For us, everything is about a positive-sum thinking game. We don’t think about how we fight within the greenhouse industry? We think the success looks like increasing the GDP of the greenhouse industry. Fresh local produce is not going anywhere. It’s growing every single year.

The only thing that’s keeping us from that is managing the risk, right? And there’s no such thing as crop insurance for growers today inside the greenhouse industry. So how do I increase my production, knowing that I’m taking a higher risk?

Meanwhile, over 60% of the tomatoes that we eat in the supermarket today are grown in greenhouses. Over 30% of cucumbers, over 30% of peppers. So we know the world’s demanding it, but we’re not adapting fast enough to it. Success, looks like increasing the GDP consistently over the next 10 years. So that way everybody succeeds and everyone gets fresh local produce. That’s what we’re focused on.

Chuck: That’s phenomenal. Nice. And so for farmers listening, what’s one piece of advice you’d give them about approaching AI and how they might work with it in agriculture?

Adam: For farmers, we recommend aligning incentives and make sure that success is essentially guaranteed for you. We believe that the way you look at an operating model and the way you look at adopting AI has everything to do with how successful you’ll be.

So taking a skeptical approach will always make it a slower adoption than taking an enthusiastic approach. And so it’s not about farmers caring about AI, it’s about farmers caring about the outcome. And AI is just a tool. And if someone can tell you, they can give you that outcome; let them do it for you.

And because farming is such high-risk, references are always important. Talk to people who’ve also worked with these companies. But the reality is you have to be able to continuously AB test and continuously try to use new tools to be able to you margin efficiencies otherwise you run the risk of falling behind.

The last thing I’ll leave you with is when you think about AI and you think about data as a farmer, one of the most important things to remember is that the beginning of time for your experience as a grower, as an agronomist, is when you started to be on the farm and learn. But the beginning of time for your AI that you turn on will be when you first integrate and put that data together. So it cannot go back and look at what happened to you 30 years ago, because it doesn’t remember that. And so the sooner you do that, the more optimization and the more value you can bring for your farm over time. And so there’s no excuse to not start now to ensure the beginning of time for whatever AI model you use in the future starts today.

Chuck: Shail, what’s your advice for farmers approaching AI?

Shail: My advice for farmers listening and thinking about approaching AI is start small. Start now. Don’t wait for the perfect data or the perfect tools. Pick one decision that costs you time, costs you money, or sleep. And see how AI can help you make that decision better and faster.

You don’t need to understand the math. You just need to understand the value. And the farmers who win with AI aren’t the ones who know the most tech. They’re the ones who learn the fastest. So treat AI like a new hire, train it, test it, make it earn your trust. And just remember, AI doesn’t replace your experience. It compounds it.

Chuck: Yeah, I fully agree and I do feel like everyone needs to be using it and experimenting with it and trying it. Yes, I’m certainly biased in the tech space, but at the same time, is a much more approachable technology than any we’ve had ever, I think from that standpoint.

Adam, Shail, thank you both tremendously. Really appreciate your time today. It’s a fabulous conversation. I’m learning so much and seeing how different aspects of our industry are starting to embrace it. And it’s really exciting to see and I think there’s going to be some fantastic opportunities for both of your firms, obviously, but also for your customers as well.

So thank you both.

Shail: Thank you

Adam: Thanks for having us, Chuck.