প্রতিলিপি 🔗
Teddy Bekele (00:08):
I am Teddy Bekele. I'm the chief technology officer at Land O'Lakes. So Land O'Lakes is a farm to fork cooperative. It's farmer-owned cooperative. So $16 billion in revenue, and we span across three major lines of business. We have our Winfield United business, which is a crop inputs business. We buy seed and crop protection from basic manufacturers, and then we sell those to our retail owners, uh, that are part of the cooperative system. Uh, we have our animal nutrition business, uh, where we make feeding formulations for every species of animals other than dog and cat, and that goes under the Purina name. And then lastly, our dairy foods business, which carries the Land O'Lakes name, which most people know us for. We buy milk from our cooperative members. We turn that into value added products like cheese and butter, of course, that you find in the grocery stores.
Teddy Bekele (00:56):
Uh, and that's been the, the, the, the origins of the company for the last 105 years. And of course, we say farm to fork because at the, in the sort of the sequence that I just outlined, uh, we start from the seed that goes into the ground, that comes out of the ground, and then we feed the animals and our number one species that we feed our cows. And of course, those cows then produce milk and butter and cheese. So we sort of have that full visibility from all the way the seed goes in the ground till somebody buys our products in the store.
Jennifer Strong (01:24):
So Land O'Lakes went nearly a century without a chief technology officer, right? So then you became the very first. Can you tell us what this job looks like from just a practical day-to-day and also maybe a bit about how that shift took place?
Teddy Bekele (01:38):
Yeah, absolutely. So we've always had more of a CIO in place, right? So, but more, that was more running the day-to-day business. So if you think of the applications that, you know, sort of run our manufacturing locations that do our financials, uh, some of the web applications, we've always had somebody do that. And that is part of my responsibility today as well. But in 2018, our CEO, Beth Ford, decided that she wanted to have technology not just be a sort of a support function, uh, but also truly a strategic enabler and then eventually a driver of some of our new innovation. And so she created this position of a chief technology officer. So in, in addition to the typical responsibilities you'd see as a IT leader within an organization, she also wanted somebody to help with digital transformation. So how do we think about some of these newer capabilities and technologies that are coming out and how do we apply those to, to business models and how do we serve first start to shift our business to be more digital in nature, but then really how do we bring more of those digital insights and incorporate them in our day-to-day work?
Teddy Bekele (02:42):
The other part of my job that got included was also the cybersecurity and privacy, uh, responsibilities. So I have those responsibilities as well. And the idea was as we go digital and we open up new doors, we open up new opportunities for, uh, threat actors to come after our organization. So how do we make sure we have a holistic view around that? And then something that got sort of, I guess was part of the responsibilities before, but it was sort of more in the background, more as like some up and coming digital capability was really this AI transformation. And that's actually where I spend the bulk of my time now is there. And obviously, we've always done advanced analytics and we've always dabbled with algorithms and different ways of bringing new insights to market. But of course with generative AI, everything just got in hyper, uh, just at a velocity that we've not seen before.
Teddy Bekele (03:36):
So really part of my work now is not just to work on some of these newer models, but it's really how do we transform the organization? How do we make sure that our entire enterprise embraces AI and we start to use it on a day-to-day basis to not only automate some of the work that we're doing today, uh, become a little smarter, more productive and faster, but how do we think of it as somewhere as a catalyst for innovation?
Jennifer Strong (04:02):
I'd love to talk a bit more about the AI part of this because as you said, most people know Land O'Lakes from butter, right? So what would surprise them most about what your team is trying to build right now or what you've built?
Teddy Bekele (04:14):
Yeah. So, you know, we have AI activities happening all across, uh, our organization. Like I said, one area is just basically using AI, and that's most organizations are doing it today, right? So, you know, how do our employees use it to be more productive, not just writing emails faster or more elegant emails or recording meetings, which that's a, that's a given, but also think about, you know, some of the work that they're doing today and automating some of that. We have a lot of pallets, for example, we will ship out to a lot of our customers. And so can we use computer vision and ways to, uh, identify areas where before the pallet goes out, we can identify where things have been damaged and rectify those before it reaches a customer and we have more issues. So normal, you know, employees are like learning how to do this, right?
Teddy Bekele (05:00):
Like build a simple agent to be able to leverage some computer vision and aid them in the work that normally would have been done through human eyes and you wouldn't have caught things as fast as we do today. So that's just more of the basic productive and teaching them how to build an agent, uh, make sure that it works properly, work in co- coordination with one another and, uh, and then embed that agent in their day-to-day work. So that's, I think, uh, most organizations are doing this today, so no surprise there. But some of the areas that others may
Load full transcript
find interesting, in our animal nutrition business, in the Purina business, we pride ourselves in doing a lot of research around what's the right feeding formulation for a horse, for example, right? How does it give them optimal performance? Uh, how does it give them optimal health for the horse owner?
Teddy Bekele (05:45):
And of course, we do a lot of research and then the, the fruits of that research are in the bag of feed that you would buy at a store for your horse. Well, now a lot of that research co- contains a lot of insights that a horse owner may wanna know more about. And so we're building these agents, for example, we have an agent called the Checkerboard Connect, which helps, uh, horse owners first, uh, better identify how to optimize the health of their horse. And a lot of that research data that we have is encompassed in this agent that we make available to a lot of nutritionists today that can look at, you know, like, hey, can you take a sample of the horse's blood sample, a stool sample, for example, and then we can quickly analyze it and turn around and say, you know, if you had more macro ingredients like corn or soybeans or, or vitamins like supplements, this would help the performance of this specific horse at this moment in time.
Teddy Bekele (06:39):
And so we're using a lot of that data and information that we have before to deliver updated insights. So that's a good way to be able to do something that's differentiated that most people don't think about. And the crop inputs business is another really interesting one. And we've always helped farmers with insights around, you know, what seed do I plant, when, and what type of soil, and what are the characteristic- characteristics of this seed that makes it special? And a lot of this insight may have been in some sort of a mobile app, could have been in a book, which we have a lot of those. And now we've actually created a super agent where we've taken information that's in a book, for example, and made it available that you can actually chat with the book essentially and say, "I'm having this type of a drought situation, and I'm seeing these types of diseases, and I think this might be the issue.
Teddy Bekele (07:28):
What are some of the signs that, you know, things are progressing worse?" And then they would get an answer back, and then sort of an agronomist now can have a full-blown conversation with this book and going back and forth. And of course, they're using a lot of their insights and experience and, and knowledge, but it's bringing them sort of that, that, those tidbits sort of at their fingertips in a much more conversational way.
Jennifer Strong (07:47):
I feel like ag tech is a story that's probably one of the great stories of our time that most of us miss. And you think about the challenges of anything from mud to internet, you know, just being much greater than they are for people in other sectors perhaps. And there's also the fact that, you know, this is
the livelihood and this is the family land for many people. And so mistakes have different types of costs than an investment of another type might.
Teddy Bekele (08:11):
That's a, that's a, that's you, you, you've na - Yeah, you, you've nailed that, right? And this is where you cannot take the human out of the, you know, I, I don't even like saying human in the loop because that sounds like there's a loop and then a human is some circle or some ring inside of it, right? Really, you have to design it with like the human at the center, right? And, and that human for us, uh, i- i- is the farmer and it's also the agronomist. Agriculture's been, you know, it, it's been, and still to this day, it's done with a lot of intuition and experience, right? So the combination of those two things. Experience being, you know, you, you've worked this land for a number of years, you know where the trouble spots are, you know what happens when it's a year that's too wet because you got too much rain or a year that has ha - heavy drought.
Teddy Bekele (08:56):
Like you sort of have seen that over the years, right? So you know how to react to it and you've learned through, you know, trials and tribulations on how to make things better. And then, and then the intuition goes into like, because I've seen it and because I've sort of mastered my recipe, like I, I know how to react to certain things. However, I don't know everything. And there's like a, a million decisions I have to make every single day. And so this agronomist is a person that is dedicated to knowing agronomy and the, and the, and the sort of the lifecycle of a plant and how to best treat it when it has certain diseases or when I'm running into issues where there's too much water. And, and I rely on that person's input to help me make, make decisions into exactly what you said, my livelihood is at stake because one mistake, and you can wipe out the entire crop for that year.
Teddy Bekele (09:42):
And every year is the year I could lose the entire farm that's been handed from generation to generation down to me and I'm gonna be the one that messes it up, right? And of course, the last thing you want is that you listen to some AI advice that wasn't grounded in the experience and the local knowledge and all of a sudden the things have gone out the window. So what we've done is, you know, how can we make sure that we build our AI tools around that agronomist and around that farmer? We wanna make sure that they bring their knowledge, their expertise, and their intuition, and we augment that. So we let them ask the proper question, and then these are the things that maybe brings them the, instead of flipping through a book and being like, "Yeah, I think that was on page 367," which they wouldn't remember that, right?
Teddy Bekele (10:26):
It's now how do we bring that out quickly so that you can make a decision? Or, but yet make a suggestion, but then s - ask for more inputs, that way there's a back and forth interaction that happens with that. So we wanna make sure that the AI is a companion, the human is at the center of it, and then we are designing, um, the capabilities around that.
Jennifer Strong (10:52):
Can you talk about a project that didn't work and what you learned from it?
Teddy Bekele (10:56):
You know, um, well, more so than didn't work, there's a lot of things we try, right, and just see how, how it goes. So as an example, like, one of the things you could do, for example, on Land O'Lakes.com, like, if you actually went to Land O'Lakes.com, you would not find our organization, like, that actually is Land O'LakesInc.com. Land O'Lakes.com is a recipe site. So if you go there, part of the way, and we try, we try to, to, uh, teach you all the novel ways in which you could use butter and cheese into your everyday cooking, right? So, uh, particularly if you're, like, during Thanksgiving or during Christmas, we have lots of interesting recipes there. So we were trying to figure out, you know, is there a way to use AI to be able to help with that, right? And, you know, if you told us, for example, "Here are the things I have at home, here's the recipe that best fits you." So, again, we were, we're trying to do it.
Teddy Bekele (11:49):
We're still in experimentation mode, but people like to just browse. Like, whether I have the ingredients at home or not, I wanna browse different recipes, and I just wanna get some ideas. So, instead of you telling me exactly what I want or what I might need based on the things that I have, I just wanna be able to browse and see, and then get inspired to make something. Um, so that was a learning. That was an interesting one to say, "Well, we shouldn't just put an AI first and say, what do you wanna make? 'Cause half the reason I'm here is because I don't know what I wanna make." Now, the key is to figure out how to have you have an experience where you're still browsing and learning and getting inspired without getting in the way and try to be too prescriptive about what you might want or might wanna do.
Teddy Bekele (12:29):
So, this is something we're messing around with and trying with, but still trying to find the right places to put those things into place. You know, what's interesting as a technology organization is we are trying to, uh, of course, like, if you come to us and say, "I have this phenomenal idea, and I wanna be able to rethink how we reshape our receiving process in our, in our sort of milk facilities, right?" We will absolutely help you figure out how to reshape that, uh, redo the workflow, redesign it, embed agents into it, et cetera, all those kind of fun things. And, and we can bring in lots of technology expertise, we can build models, but part of where I find our organization is shifting to is becoming less of the doing side of it and more of the educating, right? Okay, let me... A lot of these tools are gonna be available at your fingertips now.
Teddy Bekele (13:18):
Now,
when you get into the more complex stuff, we're happy to help and get embedded, and we will certainly help you with this, but the best person that can redesign the workflow is you yourself, right? And again, and this also takes away this fear of, like, my job's getting replaced because, like, no, no, no, it's your job to redesign the workflow, figure out where the agents fit in, and then redesign your work and how you think about this going forward. And so we try to do a lot of education and a lot of, uh, teaching about how we can use these capabilities to be able to change, uh, sort of the day-to-day work that we do all together. So that's an important aspect that's fi - we find ourselves in a different place than maybe we were five or six years ago where you came to us, asked for something, and then we delivered on those promises or those, or those other things you asked us to, uh, to deliver.
Teddy Bekele (14:05):
So interesting how the whole organization is changing, and so are we in the process of that. The, the other piece that I think I, I'm finding is, is one of the greatest things for us is this farm to fork visibility, right? So as much as you wanna be producing things that are what is needed, right, like, what do you need to have in this one given store? Like, how much quantity should be produced for that store? Or, you know, for the type of animals we feed, well, how much raw mat - raw ingredients, like corn and soybeans, for example, do you need in the given month, right? And what you find is in supply chains, it's more of a push than it is a pull, right? I mean, you wanted to make it as pull and demand-driven as much as you can, but because you don't know what the demand is, you end up being more of a push.
Teddy Bekele (14:52):
And what I'm finding with AI is we are shifting more and more to demand-driven. You could start to forecast based on history, based on sort of what latest variables are changing in the market or with the given stakeholder group, like, what might be the demand? And so then you start to produce more for what's needed than just pushing it down, right? And I mean, and, and that's a tricky balance, because if you think about milk, for example, for us, I mean, you don't tell the cows, "You know what? We're gonna milk less this month because the demand is lower." The cows, they, they show up at the milk, at, at the milking station, and then you milk them, and then you have the milk, and you gotta figure out what to do with it, right? So whether you like it or not, by nature, you're sort of forced to do that.
Teddy Bekele (15:38):
But you could do things like, you know, how much of a herd do we need at any given time, knowing the, what the weather is going to do, because depending on whether it's hotter or cooler, there may be more or less production of milk. And then we do know that demand at Thanksgiving and Christmas is much higher, uh, but again, we can't just tell the cows to, "Hey, it's, it's time's coming. It's, you know, it's, it's now the holidays, you gotta produce more." So we gotta figure out how to optimize that whole supply chain to be reacting to these types of fluctuations. So that is something that we've managed sort of by experience a little bit over the years, but now AI is really opening up more the ability to control that better and actually resolve some of the, more of the issues we've had in the past.
Teddy Bekele (16:22):
So that's interesting to see in how AI is starting to shift the entire big supply chains, not a little small contained ones, but very much wide across the band.
Jennifer Strong (16:32):
This has been super interesting. Thank you so much, Teddy, for taking the time to chat. You're
Teddy Bekele (16:36):
Welcome. Absolutely.