This week’s SFTW ConvoTM features my conversation with Greg Chiocco, VP of Engineering and Product at Farmwise.
Greg excels at the intersection of hardware and software, bringing complicated products to life that provide real and immediate value to customers. He loves spending time around cars, and industrial equipment. He has spent a career spanning more than 20 years working at different Ag, AgTech, and consumer hardware companies like Mineral, Granular, Zoox, Climate Corp, Corteva and Trimble. His last stint was with Farmwise. He also has experience working in software, strategy, commercialization and product management.
Greg could have been a farmer in another life. He is a straight shooter, loves tinkering with hardware, and does not take himself too seriously. He often participates in amateur car racing. He loves cars so much, he named one of his kids after a famous car guy!
Given his wide variety of experiences in different parts of Ag and AgTech, he has a ton of experience in terms of what works, and most importantly what doesn’t. He has the scars to show it.
In this conversation, we dug more into robotics, operations, how to sell to growers, and the massive importance of logistics and distribution. I hope you enjoy and learn from this SFTW ConvoTM, as much as I did having the SFTW ConvoTM.
Greg Chiocco, ex-VP of Farmwise. Image provided by Greg Chiocco (Artwork by EI)
Summary of the conversation
This conversation centers around agriculture robotics, particularly the challenges and misconceptions surrounding their adoption. Greg Chiocco discusses defining "robotics," overcoming farmer skepticism, and building a successful business model for ag tech. He emphasizes the importance of demonstrating tangible value, addressing logistical issues, and providing robust support and training alongside new technology. Case studies alone aren't enough; a portfolio of evidence and practical, hands-on demonstrations are crucial for farmer buy-in. He also touches on the future of ag robotics, the role of data, and the potential for an incubator-style model to aid startup distribution.
Build Less, Scale Faster: The Rise of White-Label Platforms
Ag leaders aren’t asking “Can we build it?”—they’re asking “Should we?” White-labeled platforms are helping agribusinesses move faster and deliver on strategy while staying focused. From carbon tracking to digital agronomy, e-commerce, and lab insights, leaders are choosing to scale through their brand, not by building new tech stacks.
Customers don’t care who built it—they care that it works. With white-labeled tech, you can go to market fast with your logo, your workflows, and our managed infrastructure behind the scenes.
SFTW ConvoTM with Greg Chiocco
Robotics and the automation spectrum
Rhishi Pethe: When you think about ag robotics, what are some of the things that people get right and what are some of the things people don't? What are misconceptions that the industry has about ag robotics?
Greg Chiocco: It really comes down to how you define "robotics." If you see robotics as level-three autonomous vehicles, then no, agriculture hasn’t seen massive adoption. But if you count semi-autonomous guidance systems, rate control on sprayers, or yield monitors on combines, then elements of robotics have been part of farming for decades.
That’s a misconception I’d call out. Same with the idea that farmers are slow to adopt technology. If you spend any time on farms, you’ll see they use a lot of sophisticated tech, it’s just more specialized and less obvious to outsiders.
Today, there’s technology everywhere on a farm, you just have to know what you’re looking at.
Rhishi Pethe: I guess we define robotics by the cultural norms we’re used to. Maybe “norm” isn’t the perfect word, but how do most people understand a robot? They watch movies, Terminator, or whatever, and that shapes their idea of what a robot is. That’s where the perception starts.
Greg Chiocco: Maybe it’s better to think about it using the common framework of robotics, the three D’s. (Dull, Dirty, Dangerous) You usually have to check all three boxes to call something a robot. But if you check one or two, you’re still part of the way there, and that’s definitely happening.
Take a modern tractor, for example. You can automate the entire U-turn sequence. That’s where a lot of the action happens, when the tractor turns, a million things are going on. You can set it up to repeat that pattern automatically, even triggering it based on GPS position.
So there’s already a ton of technology and automation happening. It might not meet the strict definition of robotics, but that’s a pretty common misconception.
Rhishi Pethe: So automation falls somewhere on the spectrum of what you call robotics.
Greg Chiocco: Maybe it ties back to what I said about the three D’s, but there’s probably another framework people use too, something that involves sensing and reacting in real time to the environment. Maybe that’s another key element of robotics. Even with that definition, though, you’ll find technology on combines that adjusts ground speed based on the density of the incoming crop, ie slowing down so they don’t jam up. That kind of automation has been around for years.
Five levels of automation defined by Case IH (Image Source: Case IH)
Rhishi Pethe: I was looking at The Mixing Bowl’s crop-robotics landscape map. The 2024 edition lists more than 300 companies. My hunch is that your definition runs a bit broader than theirs, but I’ll confirm. You’ve worked at FarmWise and at Zoox, which, in its own way, is a robotics outfit.
So, thinking about the work you’re doing at Zoox and what you did at FarmWise, where do the challenges line up, both technologically and in getting customers on board?
Greg Chiocco: That’s an interesting parallel. I’ve never really done that before, I always thought of them as two separate worlds. When I was at Zoox it was around 2018, they were pretty far from market entry; they were deep in the prototype phase.
Maybe I’ll break it down from the technology side. What impressed me most at Zoox was their ironclad model improvement process. They had a really strong system for collecting data, identifying performance issues, quantifying the challenge, and closing the gap quickly. They organized their teams into buckets, one team focused on road markings, another on road signs, another on rider quality, and so on. They focused on key performance areas and iterated really fast.
When I moved to FarmWise, it felt similar but different. At FarmWise, the technology had matured a few years and they ran a very clean MLOps pipeline . Granted, it was a slightly simpler use case, but they also leveraged a foundation model. That meant we could fix corner cases, squash bugs, and boost performance in a matter of hours or days. And onboarding a new crop, which would have been a huge lift before, took just two to four weeks.
So there were lots of similarities in that tight, closed-loop iteration, it was just fascinating to see how much the technology had evolved and how differently it was applied at FarmWise.
Startup challenges for early adoption
Rhishi Pethe: It sounds like the overall process stayed the same, but when you got to FarmWise the tools at your disposal were more sophisticated. That upgrade lets you squash bugs, roll out fixes, and add new capabilities, like support for another crop, much faster. Still, all that tech only matters if the farmer can say, “Yep, this actually solves my problem.” So, what problems does a company like FarmWise tackle, and how do you build a solid business case for each buyer, whether it’s an individual grower, a farm manager, or a big outfit like Taylor Farms?
Greg Chiocco: This all starts with ROI. From what I’ve seen over the years with tech adoption on farms, most farmers want to see some kind of return when they apply a new technology. If it’s a big capital purchase, they typically want ROI in two years or less. Some farmers might tolerate a little longer; some want it even faster.
Agriculture is unique in a lot of ways when it comes to quantifying ROI, and this can go in a lot of directions, so keep me honest here. For one, many farms work on a one-year crop cycle, which means you only get one chance each year to measure the real impact. But in veggies, and this is why a lot of startups focus on berries, lettuce, things like that, you get multiple turns per year. You can measure outcomes faster, and the crops are high value.
Another path is the intangibles. ROI isn’t just a clean number on a spreadsheet. When you bring technology into a farm operation, a lot of value hides in places that are harder to quantify.
After visiting hundreds of farms around the world, I can tell you that most farmers aren’t all that concerned about agronomy, they have agronomists on staff, they understand their soil and crops really well. What kills them is logistics. They want to know: how can you make my life easier logistically?
That’s where intangibles come in. Take something like the Vulcan weeding robot. Sure, you can set up an ROI model: here’s how much you’re paying for hand crews or spraying or whatever you’re currently doing, versus the cost of buying and running the robot. You can depreciate the machine and make the math work.
But the real, game-changing value often comes from something harder to measure: just being able to send a robot out instead of scrambling to find a hand crew. Knowing it’ll work in tough conditions day or night. Knowing it’ll show up no matter what.
I saw the same thing years ago with the guidance products that precisely steered tractors using GPS. We sold ROI at first, skips, overlaps, and efficiency . But what really won people over was the experience: being able to sit in a machine and not have to worry about steering. Farmers could focus on running the implement, or check their commodity prices, pay a few bills all while sitting in the can, basically get a couple of hours of their day back. They could even put a hired hand in the planter tractor and know it would be okay, work at night, work in tougher conditions. That’s what sold the technology, not the spreadsheet math.
One more thing, and this varies widely: even at the most sophisticated farms, whether it’s a massive multi-million-hectare operation in Brazil or a one-acre plot in Germany, farm accounting usually focuses on commodities, not on detailed operational tracking.
So when a farm tries to calculate the ROI on a specific robot doing a specific job in a specific field, they often realize they don’t have the resolution in their data to get a clean answer. They are focused on the big picture and there are too many confounding factors in real-world farming to precisely quantify the payback they are getting on a technology purchase.
One anecdote: we had a grower try to run a full cost analysis after using our robot for a season. They realized halfway through that their own numbers weren’t good enough to really tell the story. The spreadsheet didn’t look great, but when you walked the field, it was obviously creating value. They loved the machine. It was doing the job. You could see it with your own eyes.
So what we’re finding is that yes, our robot is a weeding robot, but it’s also a tool for exposing just how challenging farm accounting practices can be.
Vulcan Weeding Robot by Farmwise (Image provided by Farmwise)
Rhishi Pethe: What do you mean by logistics?
Greg Chiocco: When you think about a vertically integrated farming operation, someone who grows, processes, and brings their product to market, about 50% of their product cost usually comes from farming itself. And given the level of automation in processing facilities, I’d guess that most of their personnel are also tied up in the farming operation, especially when you’re dealing with fresh vegetables where there’s so much hand labor involved.
When you manage something like that, you have to drive a certain tonnage into the processing plant every day to keep it running at optimal efficiency. That puts a huge amount of pressure on the farming side. And since most farming happens outdoors, there are many confounding factors you have to deal with. The bigger the operation gets, the more complicated the logistics become.
I’m talking about things like this: if it rains one day in Hollister, and that’s where all your equipment and crews are staged, and your crops are maturing, now what? You have to figure out how to move your harvest crew. Maybe you put them on buses and ship them two or three hours south to King City. You’re scrambling to adjust, and that kind of thing happens every single day.
It’s just a massive amount of overhead. I wouldn’t call it waste, it’s unavoidable. But it’s a huge logistical burden tied to the uncertainty of running a farm. And that’s what I mean when I talk about logistics: the daily complexity and overhead that comes with trying to keep an operation moving.
Winning the first customers
Rhishi Pethe: Take FarmWise as an example, what technologies were you actually putting in the field? You might estimate ROI with only so much confidence, but as a robot developer, product manager, or operator of a company like FarmWise, what do you do when you show up on the farm?
How do you build the goal of keeping logistics unchanged, or, better yet, making them easier and less stressful, into your product decisions and go-to-market plan?
Greg Chiocco: The way I see it, and I think FarmWise did a really good job with this, is that tech adoption on the farm usually goes one of two ways. Sometimes the folks on the ground believe in the technology, want it to make their lives easier, and have to convince leadership to make the capital purchase. When that happens, things usually go well because the people who will use it are committed to making it work.
But other times, leadership at the top says, "This is the future, we need to get on board," and they purchase the technology without fully involving the farm managers in the decision making process. . Then the folks on the ground are like, "I already have a system that works, and now I have to deal with this new thing."
What FarmWise did really well was handling that gap. We showed up with the machine and an implementation specialist, an expert user and trainer. We kept that specialist with the machine for at least a week, sometimes two, helping the farm team integrate it into their workflow and handle logistics.
There was a focus on high quality user training. Sometimes our specialist operated the machine while the farm's operator watched. Other times, we consulted with farm managers, “Hey, it might work better over here,” or, “Give us your planting dates so we can help you optimize when and where to run the machine.” That's where we saw the most success.
We had situations where a machine showed up, and the farm manager was like, "What is this thing? Get it out of here." We saw some early resistance, even a little sabotage, like not moving irrigation pipes to block us from entering a field. But over time, once they saw the machine’s value, those same people became champions. They'd go back to leadership and say, "I want more. Get me another one." We converted a lot of naysayers that way.
But the bigger question is: how do you even earn the right to be on the farm in the first place? You can’t just calculate an ROI upfront and expect them to buy. They have to experience it. They have to see that it doesn’t complicate their logistics or operations.
To even get there, you need a few things like a realistic ROI calculator, clear communication of intangibles, demonstration at other farms and on-farm trial opportunities. And that ties directly into the case study question you asked earlier, we're skirting right around it. Case studies become crucial once you have that early validation in real farm conditions.
Rhishi Pethe: From a sales angle, you might start by telling growers, “You don’t have to buy the machine right away, we’ll deliver it to your farm, let you run it for a few days or weeks, and send an implementation specialist who already knows how to operate it. We’ll work with your team, fold the machine into your current workflow, and show you exactly how it helps.”
I’m guessing you ran that demo a few times before farmers could say, “I saw it working on my neighbor’s field, I trust them, and the results are clear, sign me up.” For a startup, staging all those demos is tough, so what framework or tactics can you, or any young robotics company, use to push past that hurdle?
How do you actually prove your claims? I’ve already heard that same pitch from twenty other companies. I’m skeptical. Why should I believe your case study?
Greg Chiocco: Exactly. I think the same thing applies to case studies. Just like the definition of robotics is fuzzy, the definition of a case study is pretty fuzzy too.
When I think about how farmers actually use case studies, and I’ve talked to hundreds of farmers over the years, I’ve never seen a farmer buy a piece of technology based on a case study alone. It comes down to a basic truth: every farmer’s field and every farming operation is unique. No matter how good your case study is, no matter how statistically significant you think it might be, and by the way, most case studies aren’t anywhere close to rigorous or repeatable, it’s never enough to convince a farmer to make a purchase.
At best, a case study serves to inform a farmer. It gets them interested in the technology, maybe curious enough to take a closer look. But from what I’ve seen as a product leader, a case study has never been the key piece of information that drives a farmer’s decision to buy.
Farmwise Vulcan being tested in Yuma (Image Source: Greg Chiocco’s LinkedIn post)
Rhishi Pethe: I doubt any single piece of information will make a farmer say, “Okay, I’m buying.” You need to build a portfolio of evidence. One part is the case study: we show up, run the machine on your farm for two weeks, log the results, and offer you a good deal that factors in all the moving parts.
A case study costs time and money. You have to do the work, gather the data, and package it, but it still carries weight. It sparks buzz, raises awareness, and adds to the proof stack. The ROI may be hard to pin down, yet the exercise isn’t useless; it helps people notice you and see real-world results. Does that sound fair?
Greg Chiocco: I think it is. And you hit on an interesting point, you really do need to put together a full portfolio that covers all these elements to convince people.
If you break it down, a case study can be useful. But you also have to market your product and generate awareness. You need to show up with an ROI calculator, a way to demo the machine in the field, and a way for customers to see the machine working and being adopted by others.
All of those pieces have to come together. And even then, you still need a huge top of the funnel. Because ultimately, you're working against a take rate, how many people actually move through that full portfolio of five or six touchpoints and make a purchase.
And for startups, that's incredibly hard. It’s one of the biggest challenges they face.
Marketing, distribution and proof points
Rhishi Pethe: Case studies matter, but I can’t run one until I have a robot that actually works in the field. So how do I reach that point? Programs like Reservoir Farms give startups test plots and hands-on farm access, yet a successful ecosystem needs more.
What other resources and support let a young company keep advancing, prototyping, iterating, and validating, until I’m ready to ship a machine to Taylor Farms, run it for two weeks, and prove its value? What intermediate steps take me from “it works in my test row” to “your team can rely on it,” so they’ll either buy the robot or at least give me a real shot?
Greg Chiocco: I don’t know much about Reservoir Farms and I believe they have a comprehensive model beyond just the test farm, but I think this idea could be helpful, having some dedicated ground for testing would definitely be a useful tool.
But honestly, Rhishi, even going back 20 years to my early engineering days through to today, I never had a problem finding a friendly farmer willing to help. You offer a little something like early access to the tech or a discount , and they’re usually very open, very welcoming. I never struggled to find fields for alpha or beta testing, even with pre-commercial products.
At FarmWise, same story, I could make a phone call and be on a field within a couple of hours in most cases. . And I’ve killed a lot of lettuce in the process. I mowed down plenty of corn and soybeans back in the day too. But the farmers were always open to it because they genuinely wanted to see the technology work. Farmers are good people. They do embrace new technology. They’re not rooting for you to fail.
If you’re asking what’s missing, it’s this: resources to help distribute products. What we really need are unbiased, low-cost (or free) resources to help get these technologies out into the world. That’s the real gap.
Rhishi Pethe: What does distribution mean?
Greg Chiocco: It’s a big word, and it means a lot. Distribution really covers everything, from top to bottom: marketing, sales, BD, operational logistics, everything it takes to get machines into farmers’ hands. And it even touches product development, because as you go to market, you have to localize the product too.
Help with that, help across the whole distribution chain, would be the real unlock for startups.
FarmWise is a perfect example. You had an excellent product that was clearly delivering real value to growers. But distributing it at scale was just so capital-intensive that it made it hard to truly break through.
Rhishi Pethe: Which entity would step in, and what incentive would drive them? Why would anyone invest their time and effort to serve fifty startups at no cost? Functions like marketing and BD feel central to any company.
First, would you even want to outsource that? Second, what incentive would motivate an organized group to help? If a team exists that can make the process easier, what reward would attract them?
Greg Chiocco: Would it be hard to imagine an incubator-style setup where all those services I just talked through are built right into the model?
You’d have an incubator that provides marketing, sales, BD, logistics support, even help with product localization. In return, they’d take a cut, some equity in the startup. And they’d run it in an unbiased way, offering the same resources across the board to any company coming through.
Rhishi Pethe: Are there parallels in commodity row crops? Are there no incubators? Is every company doing this again and again? Maybe there's things that could be shared.
Greg Chiocco: Totally agree. From what I’ve seen, there are a few for-profit companies that offer this kind of support. They’ll help you with marketing, BD, logistics, but it’s mostly a time-and-materials model, maybe with a small equity piece. And honestly, that setup is only marginally better than the startup trying to do it themselves.
What I’m thinking about is something different, a venture-backed model, or maybe a consortium of OEMs coming together to build it.
And if we really look, I’m sure something like this already exists somewhere. If you and I dug around, I bet we’d find examples in Europe. I think some of the OEMs there have started collaborating on programs like this.
But if you could really focus on the key components of distribution, I think it would be a huge shot in the arm for startups.
Building a startup support ecosystem
Rhishi Pethe: You mentioned marketing communications, what else? Business development? But should a third-party consortium really handle BD? Someone inside your company will represent what your product can and can’t do better than anyone else. Do you truly want an outside group running that for you?
Greg Chiocco: I get your point, no one’s going to care as much as you do, just like you’ll always be the best parent to your child. Agreed.
But to sell Vulcans, we didn’t need to move a massive number of machines. Still, even if you only need to sell 30 units, and you assume a 10% take rate, you still need 300 hot leads.
What I’m suggesting is this: yes, you absolutely need in-house BD people. But to manage 300 serious leads, you’d need 10, 20, maybe even 30 BD or salespeople. And for a startup with 40 or 60 people total, that’s just impossible.
So what I’m looking for is a force multiplier. This external group wouldn’t replace your in-house team, they would act as an unbiased force multiplier to support and scale your efforts.
Rhishi Pethe: Maybe ten startups, FarmWise, Verdant, whoever, band together and spell out exactly what kinds of leads they want. They hand that wish list to a consortium, which then hits the field, hunts down those prospects, and brings them back. The startups pay the consortium.
Meanwhile, the OEM-backed group keeps an eye on emerging tech it might eventually buy. Every week the consortium returns with another batch: “Here are ten fresh leads, all screened to your specs.” It parcels those leads out to the startups in the network.
That solves one gap: qualifying leads and feeding the top of the funnel. In most industries you handle that with plain-old outbound sales. So what else do we need? You mentioned MarCom. Are we talking about quicker access to trade media and industry pubs, or something different entirely?
Greg Chiocco: It’s all outbound, building your outbound narrative and delivering it clearly across key geographies. And for a small startup, where the marketing person already has seven other jobs, doing that well is incredibly hard.
Trying to pull that off on a global scale makes it even harder. Vulcan is absolutely a technology that could be applied worldwide. But having the ability to build and push that narrative when you’re a tiny startup is almost impossible.
I guarantee there are startups that had great products but died on the vine because they simply couldn’t get the word out.
Rhishi Pethe: Alex Rampell famously said the battle between startups and incumbents: can the startup get to distribution faster than the incumbent can get to the product?
Greg Chiocco: I’d actually challenge that a little, because I don’t think it’s a universal truth.
In some of my past experiences, working at a big company, an incumbent, you could absolutely move as fast, if not faster, than a startup, if things were set up right. No doubt about it.
At one company we operated very much in the intra-preneur model, just enough budget to operate, and the huge advantage of an existing distribution channel we could lean into. We also had the marketing, the and operations support , it was all there.
But internally, we were still operating like a startup. It was hand-to-mouth. If the product didn’t go, it would have disappeared. Simple as that.
Innovation models
Rhishi Pethe: Are you suggesting that a small, dedicated team inside a bigger company, one that already controls distribution, could really help the larger org move fast and try a bunch of things, as long as the company gives that team enough freedom and the right tools?
Greg Chiocco: Absolutely. That’s exactly what I’m saying. And I’m not arguing that one is better than the other, or that you should always choose one path over the other. It’s just that, under the right setup, a small team inside a larger company can move just as fast, or even faster, than a startup.
Rhishi Pethe: It sounded like the Alphabet X model. Here is a sequestered group of people and we give them a bunch of money. And if it works, you will get access to all the Google distribution networks.
Greg Chiocco: I very much enjoyed working at X but it is a different model than the one I just described.
First, there was no big budget. We essentially needed to make something from nothing. It was very much an internal startup mentality, but with the bonus that we had access to an existing distribution channel. That setup keeps everyone motivated and hungry. They have to deliver real results.
There was an element of the X model that encouraged partnering with a large company right out of the gate. Without a solid product/market fit you run the risk of falling into a professional services model where you essentially are paid to do a project for that specific company rather than build a product you can market widely.
Four main types of innovation models (Image source: Alcor Fund)
Rhishi Pethe: You motivate that group by saying, “Hey, if what you’re building works, we’ll back you with the resources, marketing, and distribution you need, so you can get genuinely excited about scaling whatever you’re working on.”
Greg Chiocco: We had a small group inside the bigger company (Trimble). We cobbled something together using some existing tools and some new technology, brought it to the organization, and got it to market. We picked up a little traction, and they said, “Great job, here’s a little more rope, go a little bigger.”
We found some success there too, so they gave us even more rope. It kept getting bigger. And then, when we hit a setback, when we went a little off in the wrong direction, they pulled back immediately until we course corrected.
It almost mirrored a VC model: “Here’s some money, here are some metrics. Go prove it. If you prove it and show traction, we’ll fund you again. If not, we won’t.” The difference was, we had built-in distribution, marketing, and support, all the things a normal startup has to build from scratch. That was the big advantage.
The main downside, though, is that the bigger company has to have the courage to release something they might be a little embarrassed by. Someone smarter than me once said, "If you wait to release a product that’s perfect, you’ve waited too long." In order to kick off the build, measure, learn cycle and find your product/market fit, you need to release something that might be a little rough around the edges. .
So for this model to work, the parent company has to be okay with that kind of mentality. For established companies that are worried about protecting their brand, this might not be an option.
Incentives for the distribution network
Rhishi Pethe: When we talk about distribution, a massive dealer network already moves today’s equipment, tractors and all that. You’ve even teamed up with a big player, RDO. How do they fit in, and how do you work with them? They’re obviously part of the distribution puzzle.
Greg Chiocco: I'm immensely bullish on the ability of dealerships to sell technology. RDO, for example, was a great partner for FarmWise. The only challenge was that the ag equipment marketplace is really depressed right now, and they had to focus on protecting their core business.
But if I look back at my past experience, we sold Trimble equipment through the CNH parts organization in the aftermarket. It took a while to get that machine spinning, it’s like steering an aircraft carrier. Big organization, lots of training, lots of time to get them believing in the product.
But once it got going, once that flywheel started spinning, it was a hockey stick. It took off.
So I have a lot of faith in legacy dealerships being able to do this. I also see strong players in independent precision ag dealerships. The Vantage network from Trimble is a good example. And in places like Europe, Brazil, and Australia, there are some excellent independent dealer groups that are really good at selling and supporting this kind of technology.
Farmwise and RDO partnership announcement from Sept 2024 (Image Source: Farmwise LinkedIn post)
Rhishi Pethe: How do you design the right incentive structure for them? It has to work for them, because if I’m a salesperson at a dealership, my commission drives me. So how do you craft a plan that truly motivates them?
Greg Chiocco: That’s a great question, and it really hits the hardest point.
When you come in with a new piece of technology, you’re asking the dealership to adopt a different sales motion. And that's no knock on them. Any salesperson, when they bring on a new line, has to learn how to sell it. They have to put energy into it.
So the return needs to be there for them. That’s really what it boils down to. In a reseller or distribution setup, the juice has to be worth the squeeze. Finding the right balance is key.
One way to approach it is to look at the commissions they’re already earning on the core products they’re used to selling. You need to be somewhere close to that. If your business model can’t support that, if your margins aren’t big enough, you either have to drive costs out of your product or find a different way to go to market.
Rhishi Pethe: We’ve got plenty of ways to knock out weeds, mechanical tools, chemicals, lasers, hot-oil blasts, steam, electric shock, you name it. In specialty crops, growers still lean on hand weeding, too.
So do certain factors push an operation toward one method over another? Does mechanical weeding beat lasers in some scenarios, while lasers win elsewhere because of variables X, Y, or Z? Do farm conditions, crop type, organic status, or day-to-day logistics tip the scales? Or will one approach end up grabbing the whole market, forcing you to take business away from other methods just to get your machine on the farm? Is there a winner take-all dynamic?
Greg Chiocco: That’s another universal truth I’ve picked up in agriculture: no technology delivers a silver-bullet, winner-take-all solution. In farming, there’s always more than one way to get a job done, and we could trade examples all day.
Nature’s variability is what makes agriculture special. When you grow crops in real soil, you confront endless complexity, two fields side by side can behave completely differently, even with the same crop, so you end up farming them in different ways. For that reason, I don’t expect any single approach to dominate forever.
Take weeding. In certain situations, mechanical cultivators outperform every alternative, and farmers have relied on them since the dawn of agriculture, they’ll keep doing so.
Laser weeders excel in dense crops where you’d otherwise crawl on your hands and knees to pull weeds. They shine there. But when you drop a laser weeder into a row-crop field with heavy weed pressure, physics push back: the machine needs huge energy to burn each weed and it has a fixed amount of output, so it has to slow to a crawl to burn each weed. A mechanical weeder skips that problem, it just keeps rolling at full speed.
Weeding modalities
Rhishi Pethe: Moving from specialty crops to commodities feels like stepping into a different world. Farm sizes jump way up, and per-acre margins on corn run so low they’re often negative, no wonder that shift turns into a challenge. Even inside specialty crops, lasers can shine in high-density plantings, but where, within that specialty space, do lasers fall short?
Greg Chiocco: It’s that bigger row-crop scenario, where crops are spaced into those wide bed lines with just a few lines of plants, where speed really matters.
Traditional cultivators run at two, three, sometimes four miles an hour, and farms have built their entire operations around those speeds. You have to stay in that range to be useful. If you go too slow, the crop will outgrow you before you can finish the job. As I mentioned earlier, given all the variability on the farm, having more than one way to solve a problem is important.
Rhishi Pethe: I heard an interesting comment from the Verdant Robotics guys, they follow a principle called “autonomy last.” His point was that most of the value comes from what happens behind the tractor, not from who’s sitting in it. If you add autonomy, you replace the person driving the tractor, but that doesn’t create as much value as reducing the need for human labor on the actual fieldwork, like weeding or other operations happening behind the tractor. Where do you think the bigger ROI is?
Greg Chiocco: I once heard a tractor referred to as the swiss army knife on the farm. It enables the user to complete a myriad of jobs. So automating the tractor is clearly going to be important. But on the flip side, have you ever watched a lettuce harvest crew go through a field? There are 30 people on that rig doing a variety of jobs. You’ve got individuals bending over, selecting each head of lettuce based on that day's market spec, cutting it, trimming it, topping it, tailing it, and evaluating it for quality, all in real time.
Each picker makes over 30 individual decisions just selecting a single head of lettuce. That’s where the real magic is.
It’s not the one person driving the tractor, in veggies it’s the 25 people with hoes, or in corn fields, it’s the folks driving the grain carts and semis that make up the bulk of in-field labor.
Rhishi Pethe: What are your thoughts on Carbon Robotics now moving toward this remote operation center setup? It’s a way to boost the runtime of their pretty slow machine, especially because the laser has such high dwell times.
Greg Chiocco: My take on it is pretty much directly from their materials, their website and what they’ve put out publicly. It’s hard for their users to find a second-shift crew to operate their machines, and this remote operations model is a way to address that problem.
And yeah, we're going a little off on a tangent here, but I do think tractor autonomy is absolutely part of the future. That said, to me it feels like a bit of a false choice to frame it as either automating the tractor or automating the implement. Eventually, you’ll have to do both.
Some companies are just getting ahead on one side of the scale versus the other. And I think you can already find markets for pure autonomy systems, mowing, for instance, has been a good early use case.
But over time, the whole system, tractor and implement, will evolve into being fully autonomous. You can already see the pattern if you look back at the early auto-guidance days. First, we steered the tractor. Then we started automating the equipment behind it, integrating positioning and control into the entire system with things like section control, rate control and boom height control. It’s just going to keep morphing until the whole operation runs autonomously.
And when that happens, that’s when farm data becomes truly critical. Right now, there’s decades of history showing the ups and downs of farm data efforts. But once you take people out of the field, once there aren’t operators actively seeing, feeling, and sensing what’s happening year-over-year, that's when the data will skyrocket in value.
Data play for the future of AgTech
Rhishi Pethe: When you’re not physically there, you lose that human experience, you’re not picking up on everything firsthand. The only way to know what’s really happening is through data. You end up needing a ton of rich data to replace what a human would naturally absorb subconsciously just by being in the field.
Greg Chiocco: Any farm manager can stand at the edge of their field, see just one or two percent of it, and still tell you, with five to ten percent accuracy, everything that matters about that field. That knowledge comes from decades of farming the same ground. So when you show up with a data product, you’re not starting from zero. You’re competing against a lifetime of lived experience.
The value you capture is just the thin margin, the incremental improvement over what they already know. And because all farm data comes with error bars whether it’s from satellite imagery, sensors in the ground or on a drone, that margin shrinks even further. Pure data products in agriculture don’t replace experience; they nibble around its edges.
Rhishi Pethe: If that's true, then were all these products like Climate FieldView and Granular doomed from the start? Was it always going to be hard to build a business around trading on just that little 5% difference? Or was it really just early investment into the infrastructure, betting that once automation takes off, all that data will actually become useful down the line?
Greg Chiocco: I think you nailed it in your conversation with Mike Stern, he navigated that balance beautifully.
The data play in agriculture is real. Companies like Bayer made huge investments that are clearly paying dividends relative to their core business. Today, a lot of the big data platforms, like the John Deere Operations Center, are focused on driving farm efficiency and reinforcing the value of their core products, whether that’s the green tractor or a bag of seed. And that model is going to hold true even as we move deeper into automation.
But as we pull people out of the field with the adoption of more automation, the role of data will start to shift more toward agronomic use cases. Every person walking a field today acts like a mini quality assurance manager, they see things, they catch issues, and that information flows back to the farm manager via a phone call, a text message or a conversation at the dinner table . Once you remove those people, you have to capture every nuance through data instead. That’s when the true value of agricultural data will explode. It's not that the value is zero today, but in a fully autonomous future, extracting those insights from raw data will become absolutely critical.
Rhishi Pethe: Looking at that landscape map from The Mixing Bowl and others, there were around 350 crop robotics companies listed. My contention is that number will shrink significantly over the next few years.
Image source: Mixing Bowl
Based on your definition of robotics, what’s your prediction for where the ag robotics market is headed? Are there areas on that map you see as especially promising? Which categories or companies do you think are most at risk?
Greg Chiocco: If I were to generalize, after living through three or so cycles of downturns and recoveries in ag markets, I’d say ag robotics will ebb and flow with the broader macroeconomic environment. When times are tough, we’ll see consolidation. When the market’s good, we’ll see a fresh wave of new companies targeting specific problems.
In my view, we need all of it. Agriculture has so much variability, different geographies, crop types, weather conditions, that we need many different ag robotics companies to cover all the corner cases.
Rhishi Pethe: Greg, thank you for joining this conversation and providing your insights. I am sure it will be very valuable to SFTW readers.
Additional Reading
- SFTW: Scaling Innovation Series: Farmwise (May 22, 2024)
- SFTW: Augmentation is on the road to Automation (Sept 23, 2023)
Dilepix The 6 Levels of Autonomy in Agricultural Machinery (Jan 2025)