Big Data in Agriculture: Land O’Lakes’ Teddy Bekele
21:12 Share

You may have used the phrase, bet the farm.

But if you don't work in agriculture, you might not fully appreciate what that means. On today's episode, find out how technology can support successful farm production. I'm Teddy Bucheli from Land O'Lakes and you're listening to Me, Myself and AI. Welcome to Me, Myself and AI, a podcast on artificial intelligence and business. Each episode, we introduce you to someone innovating with AI. I'm Sam Ransbotham, professor of analytics at Boston College.

I'm also the AI and business strategy guest editor at MIT Sloan Management Review.

And I'm Sherwin Kodobande, senior partner with BCG, and I co-lead BCG's AI practice in North America. Together, MIT SMR and BCG have been researching and publishing on AI for six years, interviewing hundreds of practitioners and surveying thousands of companies on what it takes to build and to deploy and scale AI capabilities and really transform the way organizations operate.

Trevor and I are talking today with Teddy Wicheli, Chief Technology Officer at Land O'Lakes. Teddy, thanks for joining us. Welcome. Thank you for having me. I'm very excited to be here. I think we first met back in 2018 when you were at Winfield United and we did a webinar together about data and analytics. Now you're at the parent company, Land O'Lakes. Can you tell us about your current role?

Yeah, absolutely. So I'm the chief technology officer here at Land O'Lakes, and it's a farmer-owned cooperative, farm-to-fork cooperative, I would say as well. We cover the whole gamut all the way from the time the seed goes into the ground, crops come out of the ground, they turn into grain, right? The grain then is used for animal feed, and Purina is also a Land O'Lakes brand, and so that grain is then used to create feed formulations for animals. And so we're

And then those dairy producers are part of the cooperative. We buy the milk from those producers and turn it into value-added products such as butter and cheese, which you find at the store. So when we talk farm to fork, you truly cover the whole spectrum here in a very large way. And my responsibility as a chief technology officer is obviously anything that runs the internal systems for our organization, enterprises for planning ERP websites, CRM, things like that.

But another exciting area is the area of data and analytics, as well as a lot of software that's used actually in our business working with our customers, whether it's supporting an existing business model or helping develop a new business model that then runs on the technology and really working with those business leaders to be able to define what that is. And then working either with the internal resources or with external third parties to be able to develop

some really neat solutions. And the composition of the co-op is what? It's really interesting. So the co-op is made up of two sides of the membership. One I mentioned a second ago, they're dairy producers, and that is how the company got started. Exactly 101 years ago, dairy producers here in Minnesota got together and said, we want to get our product out east. And so they created this co-op route to be able to handle more of an aggregation of all their products.

The other side of the membership are ag retailers. And so these are independently and operated businesses in rural communities. And they serve row crop farmers, predominantly row crop farmers in those communities. So corn, soybeans, wheat, canola, etc.

And what we do for them is as a supply cooperative, we buy the large seed crop protection products from large manufacturers and we sell it to them and then they sell it to the farmers that they serve. And in a lot of cases, these ag retailers are also cooperatives themselves. So their boards are made up of the farmers they serve. So our board is made up of half of these dairy producers and the other half is these ag retailers. And that combination is the Land O'Lakes Cooperative.

And as the CTO, what's your mandate for the co-op?

We are here to serve the members all the way down to that row crop farmer and making sure that we make them competitive at the farm gate, whatever they need to do on a day-to-day basis to stay competitive. And particularly as a chief technology officer, how do we keep up with the latest trends? How do we make sure we apply those technology trends to the solutions we give them? How do we make our supply chain more efficient? And really, I'd be here for another 100 years to make sure that we're successful.

And technology has been transforming the ag business from what I understand. Can you just give 101 of that to our audience so they get a sense of how is technology really changing the agriculture and farming business?

It's one of the oldest industries in the world. And for hundreds, maybe thousands of years, it was done exactly the same way. And in the last hundred years, you've had some tremendous changes from starting with the tractor. Now, all of a sudden, things get a little more automated. When crop protection was introduced, that was a big deal.

You did all this work, it could all be gone because of some pestilence that occurred on the field. The next piece was the biotech revolution. And so now, you know, there's a lot more research and a lot more technology that went into the seed that goes into the ground. And could it be drought resistant? Could it still come out of the ground when it's too wet? Could it be resistant to certain diseases? And that was a big game changer. And so you call it kind of the third revolution. But this fourth revolution in agriculture is

is now the application of technology, and particularly software technology, to ag. From simple things of just being able to run descriptive analytics to various components of artificial intelligence. And this fourth one, though, it's put agriculture in a hyperdrive.

Because all of a sudden, you're starting to see all this complex environment that really was managed with intuition. And now your productivity goes up. Now data and the technology is really helping you through this. So you're more and more reliant on data coming off the field and then making decisions. That really is the change in what's happening in agriculture today. And that data is really massive, right? I mean, many different sources, many different sources.

permutations, right? There are so many things that happen from, obviously you have weather, which is a big component in agriculture. In aggregate, we probably do a better job of predicting weather. But when you think on a farm, I mean, you need it down to a little three meter by three meter pixel. Like, is it going to rain in this specific spot?

Now, can you predict that with 100% accuracy? And can you tell me exactly how much rain and when is it going to happen? So that we're at that hyperlocal level, we're not quite there yet. So you have all that variability that happens. But imagine all the massive amounts of weather data you can capture that goes into sort of figuring this out. Soil composition is yet another big piece.

Then there's the selection of the right type of seed. You've got to take that into account. And then we talked about applying nutrients and applying crop protection. When's the right time? How should I do it? There's 40 macro decisions that a farmer has to make on a field, and there's 180 sub decisions. And all those could impact what happens at the end in a variety of ways with all the unpredictable climate changes and things that could happen along the way.

Really millions and billions of possibilities, right? Billions, maybe even trillions. Yes. So when you were describing this, you used the words aggregation and cooperative. I like those words. They're great ideas. They're strategic, but they're not obvious in your context.

I actually use one of your examples in class every semester when we talk about data and strategic positioning. Maybe can you describe a bit about how this testing and experimentation works? Yeah, absolutely. And we get into more of the work we're doing in artificial intelligence, which is super exciting, by the way. But part of the role of our cooperative is how do we help that farmer be more productive on their field, how to provide them better intelligence. And the way we've done it is we actually have said, because we're sort of this aggregate entity,

Let's go work with all the manufacturers and get all their seed, even in advance of when it's commercially available. We used to have about 200 applied research plots throughout the areas where we did business. And in a variety of climate types, soil types, different practices, we would plant all these seeds and then apply it on those acres and applied research plots. And then we want to be able to say, look, we've replicated this enough times to where we can say, we know exactly how this will perform in this given climate type, et cetera.

We used to have 200 of these plots. And I think, Sam, last time we talked, that was about 210 or so. We're down to actually 115 sites now versus the 250, although we cover more geography. Primarily, a lot of the work was we tried to figure out, like, okay,

what's the key component? And we got down to, it was three things. It was climate, it was soil, which no surprise, and topography was the other one. And Sam, to kind of round out your question, the idea is we do all this research, we get this information, we get these analytics and these insights, and then that's what we offer to our farmers along with these are the products you can buy, but these are the best selections because we still make it the farmer decision at the end of the day, but these are the best selections for the output you can expect at the end of the day.

This is fascinating, Teddy. And one of the things that Sam and I have researched quite a lot, and, you know, I see in my work all the time is the success of an analytics or AI program is very highly proportional to the interactivity that it allows the user or the agent or the person in charge to do it, right? So that you can understand how it works, you could agree or disagree and

AI learns something and human learns something, and that way they both get better. That's what we call organizational learning. So how does this work in the farming industry? So you provide the insights and the decisions, and the farmer agrees, fantastic, or says, no, I want to overwrite this. Is there a feedback mechanism, or are you guys thinking about that? Is that even valuable?

It absolutely is valuable. And there's one other actor in this equation that's really critical to all of this, which is this agronomist, which is almost like the personal advisor when it comes to anything related to agronomy for that farmer. A really good agronomist is a really, really strong, trusted advisor of that farmer.

And so, like I said, those 180 some sub decisions, those farmers have to meet. There's everything from not only the, some of the things we talked about, soil, climate, all that, but there is equipment, when to buy, not to buy, how to use the equipment. And when it comes to agronomy, this agronomist is the person that would present them this information.

And the reason I mentioned that is for us to be able to get validity as well as to make sure that what we're proposing makes sense and it's valuable is to make sure this advisor is bought in. And they are the feedback mechanism for us.

That is fantastic. This is so fascinating because it's such a big data play, like truly big data play. And the power of the co-op, it makes it even exponential because every experiment actually empowers the whole co-op. And I'm

I'm assuming a farmer in Minnesota isn't really competitive with a farmer in Cleveland. So the information sharing shouldn't be an issue. Like it might be in retail or some other industry, right? Yeah. Yeah. So with all these data and artificial intelligence machine learning models you're doing, somehow you've gone in this fourth revolution or fourth, I can't remember what you call it, fourth? Revolution, yes. Yeah, wave. Yep. What's next?

With some of the capabilities, both the biotechnology as well as the software technology component of it, there's farmers that can get up to 540 bushels per acre.

That's the potential that's up there. And so some of it is now is getting into this practice and is capturing data on the field all the time, making the adjustments. Some of it is just changing a lot of legacy practices. So particularly here in the Northern Plains, one of the things that people apply all the fertilization in the winter because it gets so cold and, you know, it snows and then it stays dry.

white forever, right? Till it melts. And then you have all that, when it melts, it goes down into the ground and now you have the moisture to be able to grow the crop. So traditionally, all that was applied in the fall, right? Right after you harvest, the first thing you do after you harvest is then get into tilling or maybe, maybe not. And then you do the fertilization.

Now, the idea is no, no, no. When you want to spoon feed, you actually should do that in the spring, like right before the crop goes into the ground, as well as once the crop is in the ground.

But that creates some logistical challenges, right? Because now you have to be in the field at a given time and it's twice, maybe it's three times. And there's a scheduling aspect. Who's going to do it? Is it the farmer? Is it the retailer that works with them? So it's kind of getting through that and it's learning that. But those are the things that folks are learning now and making some of those adjustments.

Yeah, that's great. You're talking about you provide these recommendations on average and then people adjust them or it's not even that they consciously maybe reject your suggestion, but they implement it slightly differently.

I always say, you know, when I started working here nine years ago, I realized when people say you bet the farm, like, do you want to bet the farm on it? People bet the farm. Like, in this business, they actually really do. Because every year, I mean, if it go badly, all the investments you make in all the crops as well as the equipment, I mean, you could totally be upside down and not be there two years later. So there's always this risk aspect of it. And part of it is they're very entrepreneurial in nature. So they want to be able to be empowered to –

make their own decision and have their own flavor to it. And that's totally okay. And all we're saying is make sure you use the data to be able to help you with that. And maybe there's a option A where you put your foot on the gas a little bit harder. And then there's an option B where it's maybe not as risky and you want to be able to balance that out. So I think that that has to be there. And farmers, they're built that way. They're built to be able to do that.

And at least lets them know if they're pressing the gas or if they're backing off. I mean, what you're doing is you're saying, here's the data that says, hey, this is an aggressive position or this is a more conservative position. Before, they were just guessing where they were on that continuum, perhaps. And now you're at least letting them know that.

Exactly. And one of the things I always tell them, you know, like if it makes you feel any better, I'm like, call it augmented intelligence. How about that? Like it's your intelligence augmented with data and some of the models we put together. It's not doing its own thing and you're not just there now reacting on what the model is telling you. So, Teddy, we have a segment where we ask you a series of rapid fire questions. So just answer with the first thing that comes to your mind. So what's your proudest moment in AI? Proudest moment in AI? Yeah.

I would say we were using this computer vision to be able to take a look at cows in the field. And could we just, with a snapshot, take a look at those cows and say, yep, it's too thin, it's too heavy, or just about the right amount of weight. But I remember when that came to life, I was ecstatic when I saw the output of that model. Perfect. What worries you about AI?

What worries me about AI is unintended consequences. You run the model and you have the best intentions in the world of reaching a certain answer, but in the background, because we don't understand what happened behind the scenes, because the machine was learning and it was writing its own code and then it learned something else. And all of a sudden we can't get to the understanding why it got that answer and we can't validate it.

But we keep going forward because it's truly too embedded into what we're doing to be able to go analyze that. So the unintended consequence of something is truly something that I worry about every day. What's your favorite activity that involves no technology?

Favorite activity of mine is Node Technology. It's trying to teach my kids how to write code. It seems like a Node Technology, but I'm trying to do it in such a way that I use Node Tech and they can have fun trying to think through the mental process to be able to do it and then say, oh, look, now you can do this.

What was the first career you wanted? What did you want to be when you grew up? A mechanical engineer. I was seven years old when I, and by the way, my degree is in mechanical engineering. So just realized I didn't quite want to be a mechanical engineer a year after I graduated. Yeah, well, I had the same thing with chemical engineering. I didn't want to do that once I graduated with it. What's your greatest wish for AI in the future?

My greatest wish for AI in the future is truly continuing the work we're doing now, is to provide help with really identify, take the guesswork. I mean, I want to make a decision as an individual, as a producer, as a farmer, as a retailer, as a large cooperative company. I want to be able to make the decisions, but I want to be able to have everything at my fingertips so I can make the most optimal decision.

So can you get me to that spot where all the information is decodified as much as possible so I can get to the best possible answer? And so I would love to take that out of the way and really get to a spot where I don't have to worry about the guesswork aspect of it. Now I can make the most optimal decision. Okay, Teddy. Great talking with you again. Thanks for taking the time. Thanks, Teddy. Thank you. Take care.

On our next episode, Sherva and I speak with Tanya Sideri, head of Novo Nordisk's AI and Analytics Center of Excellence. Please join us. Thanks for listening to Me, Myself, and AI. We believe, like you, that the conversation about AI implementation doesn't start and stop with this podcast. That's why we've created a group on LinkedIn specifically for listeners like you. It's called AI for Leaders. And if you join us, you can chat with show creators and hosts, ask your own questions, share your insights,

and gain access to valuable resources about AI implementation from MIT SMR and BCG, you can access it by visiting mitsmr.com forward slash AI for Leaders. We'll put that link in the show notes and we hope to see you there.