What if you could give AI years of your farm’s data and ask it to challenge the decisions you’re making? Joe Dykhuis, President at Dykhuis Farms in Hamilton, Michigan, took that question seriously. After Chad Fiechter challenged the idea that AI would change much about farm management, Joe decided to build his own farm-management AI tool using Purdue farm management materials and Dykhuis Farms' production and financial data. What happened next surprised Chad and Todd. Joe explains how AI challenged assumptions in a major capital investment decision, acted as an “adversarial checker,” and analyzed about 10 years of hog-loading data in minutes. It also identified patterns involving weather, cleanout timing, labor and logistics that would have taken far more time to uncover using traditional analysis. AI is only as useful as the data, context and management fundamentals behind it. Joe, Chad and Todd talk about what this means for farmers, where AI can actually save time, why “truth sources” matter, and whether the future of farm decision-making might be less about building the perfect spreadsheet and more about building your own tools as you need them. This isn't a conversation about replacing farm managers with AI. It’s about whether AI can become another tool that helps farmers ask better questions, examine their assumptions and make more informed decisions. In this episode: How Joe built an AI tool around real farm data Using AI to challenge a major capital investment decision What 10 years of hog-load data revealed Why good data and farm-management fundamentals still matter How AI can connect information that wasn't originally collected for analysis What farmers should think about before turning AI loose on their own data Subscribe to the Purdue Commercial AgCast for research-based conversations that help commercial farmers and agribusiness professionals make better management decisions.
Audio and video play straight from the show's own hosting. PodSubs stores and serves no media.
Transcript
From the show. Tap a time to play from there.
We're hyper-obsessed over tracking how many pigs that get put onto trucks and go to Tyson don't make it as full-value pigs. So they either die in transit or they aren't mobile enough to make it through their supply chain, so they have to be sorted off to the side. Every load of pigs we tag with the person that marked the pigs, the load crew leader, the truck driver, what type of truck, what type of trailer,
we load that into our system, and then we always look at driver performance and load crew performance and grower site performance Those are the three things we look at, we obsess over. So I fed that stuff all in and told it some of the biases of things we thought were important . I gave it 10 years of information. In, 15 minutes, it's analyzed it all, and then it comes back with a whole report. It came up with all the different factors and scores for which type of things have what impact. And it said, "If I normalize everyone and give them, the same load, here's the
growers I would expect to outperform the other ones." I could have never done that. I don't have weather data in, but because it knows where my sites are and it knows where we deliver pigs to, it said, " Why are you shipping pigs out of the barn, when it's less than 32 degrees? Welcome to the Purdue Commercial AgCast. I'm Todd Kuethe, and I'm joined, by my co-host… Chad Fiechter.
Hey, Chad, why don't you give us a little bit of a backstory. Our guest is Joe . We've had a couple of interactions, I think that you were one year ahead of me in ag economics at Purdue University. When did you graduate? 2007. 2007. So our paths crossed. I graduated in 2008 and then we reconnected because you came and spoke to my farm management class, and then, you reached back out because Todd and I had made some comments about AI, and you were like, "I think this thing is
a big deal." And you said, "Why don't you come up "and have a conversation Joe, do you wanna tell us more about, who you are and what the heck we're doing? Yes, I'd be happy to do that. Guys, thank you for having me. So here's what happened, Chad. You made a comment about AI that was, I felt, maybe derogatory to AI, saying that, "I don't know how this is gonna change anything in farm management." Or at least I took it that way, And I had been dabbling at that point in time,
and I thought, "You know what, Chad? Challenge accepted. Let's see what this can do." And I had been right on the cusp of breaking through, giving an AI tool some real data. So I, because I had spoken at your farm management class, and because I am a graduate of that class. I thought, "You know, I wanna make a farm management AI tool that's equally
as good to the AI tools that we're building for the people in our hog barns taking care of pigs," 'cause we had been actively loading, some information for them to be able to make better decisions in hog barns. And I thought, "Well, why should they have a cooler tool than me? With AI, I need to build something for myself." I had the thought of, you know, I, I know I sit on piles of production data and financial data, et cetera.
But how do I build a tool that makes me a great farm manager? Well, I have the world's greatest farm management course sitting right here. So why not use that to train the AI, right? So I had been thinking about that, and then you scoffed at AI, and I thought, "You know what, Chad? I just need you to send me electronic versions of the class information I will feed that into what I aspire to create the world's greatest AI
management tool, and we'll use that as the basis of, building out a tool. I gave that idea to Claude, the AI I've, I'm building this in, and It created a multi-page plan to send to the professor, which was Chad. So I sent that to you as "Give me your class file and I can do this." And you're like, "Okay, that sounds fun." So then I started loading up Dykhuis Farms information in it, and asked…
the whole point of the project was to have the AI measure what I'm doing at Dykhuis Farms against what the class says great farm management looks like, but also measure the farm management class material against a real working farm. And then I said, "Well, you can come up and we'll see what it tells us." And I think you guys left. You gave up after six and a half hours of talking about it Admittedly, at least for me, I won't speak for Chad, but I limped home.
It was a lot to take in. I feel like we drove quietly. We drove in silence for 20 minutes- Yeah … at least. Yeah. And then we'd have things that we picked up on in the conversation. Yes. There was two other people that sat in on the conversation that I think are interesting to know who else was in the room. So your dad, Bob. Yep. And Bob is still extremely engaged in what you're doing and thinking about it, and he is not an AI skeptic. Like, what I walked away is he was like, "This thing…" How old is your dad?
He turns 69, next month. I picked up no amount of skepticism. He was like, this is cool. This is really cool stuff." And then also Steve Kluemper, who used to be a banker, now works as a consultant, a farm financial consultant, correct? Yep. He's like our for-hire CFO specialist. Okay. So like the room was the, intellectual horsepower, we might have been bringing up the tail. At least I'll not speak for you, but I'll speak for myself. One of the things Joe was using the AI to look at was dealing with his
air filtration in the hog barns. And to be honest, I hadn't even really thought… I knew hogs I don't know a lot about. I'm still learning. It's not the area that I have any expertise, but it's fascinating to me. And I knew disease risk was a problem, and that you would have to shower and clean and keep a clean environment, but I didn't think about the air. You know what I mean? Yeah. Like that's how behind I was, right? So yeah, clearly I was bringing up the tail 'cause I haven't even thought about the air before. Then the other piece here is you're a little rare in the degree to which
you have been collecting data and, purposefully trying to use the data to make decisions for a long period of time. Do you feel like that's an accurate statement? Uh, absolutely. Yes. So my first job at Dykhuis Farms, my family's company, out of Purdue was my dad made up this title called farm economist, and the team was in the middle of installing a managerial accounting system.
So we have lots of sows, lots of growing groups. We raise corn and soybeans, so there's a lot of activity. And if you produce a commodity, you should wanna know what your cost of production is. And, so we were working on installing a complicated accounting system that could give us cost of production reports as well as create, inventory cost value for our balance sheet. So this is, high-level accounting, database thinking.
I was gonna work with the outputs of that system. That's hard to execute in an operation. You need to know what you're doing. So I ended up working backwards from wanting great data coming out to, "Wait, how do we actually collect, put the data in, and build a data culture to do that?" And the system we were using was called FBS Systems, which was made in the '80s by Norm Brown, who built this computerized system in the '80s around the Farm
Financial Standards Council- rules. So that's what that system was built on. He had modernized it some, and there's not been a huge number of people that use that system 'cause it's really complicated. We were implementing that, and I got way deep in the weeds on that. Fast-forward to today, we ended up buying that company with about 20 other farmers, and we're running it like a co-op. We've reinvested in it to get it up to m-modern code, and,
we've hooked AI onto it now. Like, we're, we're really doing some cool, some cool things with it. But that's where I learned about the discipline it takes to collect data, to store data, and to think like a database. In production ag, and especially in pork production, we have the benefit of every week we're doing a new replication of production, right? The whole system advances one notch. You get total outcomes. You get new things started. You get midpoint data we produce around 500,000 pigs a year.
If we can save 50 cents- Yeah … per pig, that's a big number- yeah … worth spending a little time on, right? So you need to be sure when you're gonna make a decision that the information that you're basing that off is correct, that it's reliable, and you at least know where it comes from. That was not, like, I didn't come out of Purdue thinking, "I wanna get into managerial accounting." I took the farm management accounting rather than the management level accounting.
That was a mistake, but why would I wanna do that? I just wanna work with the numbers on the backside. But that, you know, so I kinda was forced to build that intimate knowledge, and now today I kinda know where every number comes from. Hmm. Which puts me in a position to know what data to take from where, and with AIs allowing me to glue things together that we didn't have a way to glue together before. So, so yes, I think I'm probably a little rare, at least medium rare, on
the amount of data that we're sitting on. And you mentioned my dad. My dad was one of the first customers of PIC in North America, and he was one of the first people using the pigtails record system, where every sow, you kept track of every… so I've, I have, in the basement of this office, I've got production records from the '80s that he was willing to take by hand. Wow. Because you need to know your numbers, and you need to make
decisions based on numbers. So that's why he didn't seem scared at all by AI, and that was his introduction to what I was doing. I had made some comments. His office is right next to mine, and I've lofted some comments over to him like, "Yeah, I'm working on this AI thing." So he happened to be here the day you were here. I was like, "Hey, Dad, what are you doing today? You wanna come up and hear what I'm doing in Claude?" "Sure, I'll do that." But that's just how he thinks anyway. More information, the better.
Process it as much as you can. So this is just an advancement of the initiative. It's a tool to help get to where we're trying to get to either way. So he, gets the concept of it. Now, does he know how to drive it? No, I don't think so. But He's 68. He doesn't need to do that. I can load him with any information he could ever dream of now. Okay, this is perfect, right? What we came away with is that the data, the richness of your data and your
understanding of the data really drove, in my mind, it really informed the way you were building the tool and the way AI was able to give you some recommendation. I had a good idea of what I would wanna feed into it, and what I wanted it to help me process that information, pull it together. That's what I was hoping it would do, and I'd spent hours with it, like- indicating, "Here's the kind of stuff I have. What could you help me do with
this?" Before I ever revealed the data. So I really had, like, a grand plan before I ever gave it files, 'cause I was scared to. I was a little afraid of the, security of giving it intimate records like family operating agreements and credit agreements from the bank and past financials and financial forecasts and production records and benchmarking data. All, like… 'Cause if it's gonna help me, it's gotta have access to all the sensitive stuff.
Yeah. How, how did you sort through that? What pushed you over the edge where you're like, "Yeah, I'll pull the trigger on that"? Yeah. So I was at, a conference with MASA, which is the software company, now Mutual Agriculture Software Association, MASA, like the corn, right? They had spun up a secure environment for me to hook Claude right onto my financial database, like a replication of it, but that was, in a secure place on the cloud, not just on my regular machine or whatever. And we had a corporate account for Claude for that, so I was comfortable
with that, but I was afraid to pull anything back out of that environment. So we spent a ton of time in that conference talking about the stuff we were doing in AI with the data coming out of the accounting system. Then there were some technical experts there, and when one of them told me, "Yeah, it's really just as safe as any of the Microsoft platforms, and it's definitely better than Dropbox, and it's probably much better than emailing a document to someone." Wait, I use all of those tools.
That's where all of these documents live, so why wouldn't I just point… I just gotta give access to a folder to Claude And it starts learning. I'm the CEO, I'm in charge of the financials, and I'm in charge of the tech policy and IT, right? I decided I was gonna do it, and I started doing it. After that conference was when Minnesota had some sort of air traffic control strike or, technical problem, so I got grounded in Omaha for hours. I spent the whole trip, the, the whole time in the airport and on the plane,
and then in Chicago, and then the delay in Chicago, just loading up documents. That was the beginning of, Vampire Joe, I guess. I was trying to build a tool that I could query the biggest decisions that I have to make. So I loaded all that stuff up in it, and then we had a 10-year financial forecast where we're looking at capital improvements to the sow farms over the next number of years, so I work at the farm, but I have
four sisters that are also owners of Dykhouse Farms, along with my parents. Part of our ownership initiative is looking at what, capital improvements are, what everyone's expectations are, for multiple years. And if you're gonna invest a generation's, you know, make a capital improvement to set up the farm for another generation, you wanna talk about that with all the owners. So I had presented a multi-year outlay, a significant capital outlay, for
modernizing the rest of our sow farms. So putting air filtration systems on and replacing, barns of heavy equipment that's wearing out. And we had done two of these already, so it's not completely new stuff. There's a lot that goes into making that decision. There's evaluating the difference between that filtered production model and unfiltered. There's assumptions about what pricing forecasts might look like, what you think costs are going to do, all sorts of stuff. And Steve Kluemper helps me with projections.
A normal projection set has over 24 million cells in it, so it's a pretty monstrous spreadsheet, and then we rolled that out 10 years. So we're looking at this filtration decision, and so I just started asking questions. When I fed that in, Claude started asking me questions, and because it knows all of our other things, it immediately started questioning all of the assumptions. It read, without me telling it, it read all the assumptions that
we made in our financial forecast. Like, how many more pigs per sow per year are going to come out of the sow farm? It figured that out, and then it said, I've looked through the rest of your production data, and your assumption's actually wrong. You should have said it's X number instead of Y. You're being overly conservative there. Now, on your finishing downstream death loss, you were a little optimistic, you know, by 0.1% or something like that." And then it said, I don't know how to
evaluate your risk model for your health status. Do you have anything on your PRRS break rates or anything?" Which we do, so I fed it all that stuff. And we do some benchmarking with health breaks and stuff, so I fed it all that stuff. I didn't tell it anything about PRRS. It came back to me with an analysis of how often I should expect break rates, and I was expecting them to break less than what the model says that they should.
But my severity is higher than what could be analog. Now, I have a lot of friends in the industry, so I've been trying… and there's new nasty PRRS strains out there that are creating significant death loss so I did some proprietary market research that I used in my presentation to the family, and it said, "Joe, none of your numbers support this. Where are you coming up with this? This is too aggressive." But then it also said, and this is crazy. It says "You-- In your forecast, you are trying to justify the entire capital
expense on production improvement from the filtration, but you're doing a bunch of other upgrades as well. It's poor analysis to fund the whole thing based off of filtration." If the filtration can pay for it all, that's great. I don't have to come up with the rest of the stuff, but it just keeps drilling deeper and deeper and deeper. It ended up providing me analysis on, one of the biggest decisions my
generation is going to make in our system. And I continue to work with this decision, with the model. Keep asking it further questions. I feed it some more information. I challenge more assumptions. And one of the roles that it tells me it can play for me is an adversarial checker. And it absolutely, is doing that, I think on our way home, we're getting back to like what are the three things we can take away? One is that like you, you are feeding in unbelievably good data that you
are super familiar with and you already used in many processes. Two is you're doing a process that you already do. So the idea that you've already made decisions about filtration and updating barns, this is a common thing that you're doing regularly that you said, "Well, how can I use AI to make this better?" And then, the third point is, is like the degree to which you are able to go into the nuance was like I… what I took away from it is the amount of assumptions that you were testing, and the iterations that you had run through to provide information
for the decision was like you couldn't have done that as you and Steve spending a lifetime computing different scenarios. Is that a fair way that you think about it? Its ability to analyze is… crazy, right? It's reducing the cost of spending that time going that next layer and that next iteration. It's not doing anything I couldn't do myself. I can't do all the calculus that it can do. It's doing some ridiculous like quantitative statistics
stuff without even asking. One of the things it keeps doing on data is running placebo tests on timeframes 'cause I'm asking it, "Did this change in production model from here to here. Do you see any outcomes on any of our downstream data?" And it'll be like, "Well, it looks like yes. I ran placebo tests against these other timeframes, and they match up too, so it's probably… that could be misleading information." So there's times where I gotta go to another AI and ask it what math it's even doing.
So it's a little scary, but it can do some things that I couldn't do, but it also does it so fast, like I could never afford to take the time to work on this myself. An example of a thing we're hyper-obsessed over tracking over time is how many pigs that get put onto trucks and go to Tyson don't make it as full-value pigs. So they either die in transit or they aren't mobile enough to make it
through their supply chain, so they have to be sorted off to the side. Every load of pigs we have moved since, 2008, we tag with the person that marked the pigs, the load crew leader, the truck driver, what type of truck, what type of trailer, we load that into our system, and then we always look at driver performance and load crew performance and site performance, like grower site performance Those are the three things we look at, we obsess over. So I fed that stuff all in and told it some of the biases of things we
thought were important and we were trying to identify a few trouble spots and make some difference. It took, I think I gave it maybe 10 years of information. In, 15 minutes, it's analyzed it all, and then it comes back with a whole report It created, if you're a baseball fan, you know wins above replacement. Oh, yeah. It came up, it came up with a statistic where, I've kinda, figured out all of these factors, pig weight and all these different things make it more
probable that, it came up with all the different factors and scores for which type of things have what impact. Like temperature when you're loading, how heavy the pigs are, how old they are, all sorts of different things. And it said, "If I normalize everyone and give them, the same load, here's the growers I would expect to outperform the other ones." I could have never done that. We were doing what we thought was the most advanced stuff we could do. And I'm probably giving trade secrets away here, but one of the insane things
it did is it pulled weather data. I don't have weather data in, but because it knows where my sites are and it knows where we deliver pigs to, it said, why are you shipping clean-out pigs, so the last pigs out of the barn, when it's less than 32 degrees? That doubles the subs and deads when it's that cold." I thought it was gonna be like, "Don't let Larry do this." Yeah, no. He's our identified person. There, it, It definitely fl- it flagged the top 10 growers, and it gave me a cost.
You know, here's what you can, here's what you could save by improving the top 10 growers. That was part of the wins above replacement calculation. It said clean outs in cold weather are bad. You should change that. We have four days of cleaning time between batches. We can't just not ship pigs when it's cold, but there's probably some things we could do. Two nights ago, I gave it our logistics schedule, 'cause I was wondering… we tag every load with what type of trailer cleaning's required to go to it, and I
just asked it if it could see downstream effects from when we started cleaning trailers versus when they weren't. And it comes back to me, "Hey, you're cleaning out most barns on Monday mornings. I thought I told you that it is, it's bad to do that. So all winter, you're loading clean-out loads first things in the morning on Mondays," which kinda works in the routine. And it gave me a number of how many dollars that was probably costing us. So when it comes to winter, we'll be thinking about that probably.
I don't know if we'll be able to solve it, but I would have never had time to pull up the temperatures from every load. We're doing, around 60 truckloads a week. So for the last 10 years, that would take me forever just to assemble those data points. But that 15 minutes. Because I'm building this all in one project, it remembers everything it's seen before, and it actually is creating me this open items list of questions that it wants me to answer that it thinks is important, and things I need to follow up on.
So it's like 170 items long, and after… throughout and after every conversation it's saying, "Oh, hey, this information is now answered, you know, 074. I can take that off the list." It's creating a tool for, an assignment list for me- as it's learning about the farm … One of the things we talked about on the way home is, what our role should be going forward, right? So like traditionally people in our role in farm management extension would be
like, "Let's build a decision tool." Like, you would tell me like, "I'm trying to decide this." I'm like, "Oh, cool, we'll build an Excel spreadsheet that'll help you walk through that, and there'll be some things you put in and you click, and then it'll tell you do or don't do this right? And I was initially thinking like, "Well, we'll use AI to build the perfect tool." Yeah. But now I'm leaning more towards no, we're gonna use AI so everyone can build their own tool as they need it. So one of the things… Well, a huge theme that just keeps coming back is
the fundamentals are still important. You still have to understand the fundamentals. Ground truthing the conversations that I'm having with AI, super important, 'cause these things can go all over the place. And so, like, is this a, proper allocation of this theory? I don't know some of the time, right? And so understanding the basics of farm management, understanding the fundamentals
of finance, understanding strategy. So spoiler alert, somewhere online, the Purdue Ag Econ Strategic Management, there's some slides out there that it gathered up for me, and it pulled… I think it pulled some stuff from an Alan Gray topic, and that's one of the other classes I really wanna load in here. And then I pulled some stuff for Ag Econ 524, Advanced Finance. So I pulled some of that stuff in, and it starts measuring, strategic management.
So it's, "Oh, based on my analysis, it looks like Dykhuis Farms is a stage four company, where you are likely to be outgrowing your management capabilities," right? And this thing's kind of condescending to Joe, I'm not gonna lie. Um, it likes to point out my inadequacies. So it will, you know, "Huh, maybe you need to talk to Alan Gray about your lack of management capacities." And it knows that Steve is my
financial consultant, and Steve owns parts of the financial projections. When I can't answer one of its questions, it will say, "May… I add to the list that we should talk to Steve about this, 'cause you don't know. Those fundamentals, Todd, are still super important, and properly applying them is something that takes practice. And now we can use the most complicated, obscure ones too,
like in everyday decisions. One of the things we talked about there was who actually ever really gets to a net present value calculation on an investment decision, right? Like, once I know I can justify borrowing the money, let's rock and roll, right? Yeah. One of the first findings this thing had was one of our core values at Dkyhuis Farms is we will maintain a safe financial position. It hates that statement.
That is unmeasurable, and if you are making a decision about your farm, you should be calculating a net present value, therefore you need to have an internal rate of return that you expect to get on your own capital, and you can't give me a number on that. So I'm adding that to the list. You better talk with the owner group and figure out what the expectations are, because you say safe financial position, and then in your family contract, it
says you'd like to see above agriculture industry returns, and I have no idea how to evaluate that, so fix it. Two questions. One is how would you give advice to, a farming peer on how to, think through AI? You mentioned a couple things that you did to familiarize yourself. And then the second one is, where do you see it going? For you specifically, what are you gonna continue doing? I want to add truth sources to the project to tell it these are
the things we're gonna use as our collective core values right? In a mission statement. It's like, 'cause it has access to everything in the internet. I don't want some, you know, le- maybe lesser university information or, information from the '70s or- Yeah I don't know how to govern where it's learning from. But if I can point it to sources of information that we agree that that's what we're gonna use, I think that's powerful.
You used to need to be able to have numbers that you could put in a database to have like a warehouse of data. This thing can read anything that's electronic. So an effort to get everything to be electronic, to record everything, anything is background information to build context for this tool, and the tool needs context. It can read meeting minutes and PDF documents. Spreadsheets that have things typed in that you didn't even think were important
all can now, be part of the database. So I, you sent me a list of things maybe we would talk about. I gave that to Claude and asked it how to answer them, and I told it, I like the humor of what happens in AI here. So I told it, "If you don't know how to answer a question, say pass rather than guess." Well, it actually told me that I didn't know… Let's see.
You asked a question like, "How's AI gonna be used in the future?" It says, "You should pass on this. You don't, you don't know, you don't have any information on what this is gonna look like." So- Man, I think Claude's rude to you. Its role is adversarial checker to me, Chad. That's part of what we agreed on it being. So, um, it told me that some of the data, a lot of the data that I have provided it was actually not collected for analysis.
It was collected for control. Daily balance checks, logistics schedules, feed budgets. A lot of the information that I have fed it are things that we just used as like, "Oh, we have to manage this process, so we need to record what's going on. We need to measure a variance here." Everything can now be part of the database and, maybe this is where we're a little unique because we have the managerial accounting system.
Everything is always tied to a cost center or a production group. We have a numbering system around that, and that appears all throughout our data, so it's able to glue all of those things together. We have some contextual things that we did throughout the records that make it so I don't have to do a lot of interpretation for it. I feel like we've done all the fun parts, now we need to do some work. So what's your favorite non-work use of AI? … After we got done breaking down farm management.
I got two short stories. One, my first initial use of ChatGPT was around New Year's. I wanted to get some sparkling wine, right? Like, some champagne. There were some types of champagne that were apparently popular on social media, which I'm not on, but I'm told, right? So I thought, "Huh, what is this stuff?" So I put it into ChatGPT. Well, it'll give you tasting notes on stuff. So I got into a little bit of a groove of buying, tasting, and reviewing
tasting notes on champagnes around New Year's- … using ChatGPT. Works great. Then more recently, throughout the summer, we have a pool by our house and we got a pool house, and I've got three daughters and a son. They're, nine to 16 now. And after about the fourth trip through 7 Brew, I thought, "You know what? I can see what's in there, and I think if I were to buy some of the
ingredients, I can probably procure them and make some of these drinks at a lower cost of production- 100%, yes. … Than, than, you know, retail." Yeah. So I started asking ChatGPT, like, what are the most… I, I will buy up to eight syrups, right? Eight flavors. Which ones do I need for the broadest spectrum of dirty sodas, mixed drinks at home for adults or for kids, right?
It gave me a list, and I bought them. And we keep having to refill 'cause we've got them in the pool house, and they have been an absolute home run all summer with adults and kids. So then you, it… I made a project in ChatGPT of that. I gave it a picture of it- And I initially started, typing and asking, "And here's the ingredients I have." You can just take a picture of the fridge or the drink shelf, whatever your inventory is at home, just take a picture of it and say, "Here's what I got.
Give me some ideas." And it will start spitting out recipes for you. So that's by far my favorite. We all share this, that, like, Claude is our work chatbot- Yeah. … and ChatGPT is our play chatbot. I also play with Gemini at times. Okay. Sometimes I'll have ChatGPT and Gemini. I'll tell it, this one told me this. I think it might be smarter than you." Well, last Friday I had lunch with a computer science professor here. And he said, "You got it all wrong. Codex and ChatGPT is so far ahead," this was his opinion, "so far ahead
of Claude Code." And I was like, "You're the first person"… 'Cause I have been in this, world where, Claude is like my coworker, right? Like, you call it your coworker. Like- I do call it my coworker. It's literally sitting here in the room with me and we're chatting about stuff. But I use ChatGPT as, my buddy. I need to figure out a drink recipe. I call it ChadGPT. Well, that's gonna stick in my head now, Todd. Yeah. You're welcome. Yeah. I have been asking a bunch of questions initially when I started with Claude, I said, "I don't trust that you're gonna be the smartest platform for
forever. So as we build this out, I need to be able to move this." So I can apparently take everything that's in the project, 'cause it's creating boatloads of these, .md files or whatever. Oh, yeah. I should be able to just move everything over to another platform and do the same thing. But here's what I realized the other day. It's way smarter than me already, and so I'm the limitation on this thing, not it. For me to go through, the learning curve of moving it over to… It's not worth it.
So I'm just gonna hope right now that An- I- I'm confident that Anthropic will keep investing into ChatGPT and it will, again, soon be the smartest one, and they'll be going back and forth. Yeah, right. So I'm just… I got enough complexity. I'm already over my ability to operate it, so I'm sticking this farm management tool in Claude. But actually, this morning, I made my IT guy come over here and allow
ChatGPT to have access to my email and calendar, and I now have a ChatGPT, agent that's called Joe's Chief of Staff. Wow. So ChatGPT's gonna be a work buddy, too. Wow. Well, I think it's good to work with your friends. I enjoyed it. I've enjoyed it, you know? When we came, you also drove us around, kinda gave us a tour of some of your farms, also some of the area. And we drove by a chicken farm, and you mentioned avian influenza.
So you used the word AI two different ways in one conversation s- And we all got it. We were all there seamlessly. And then, I was talking to somebody earlier about artificial insemination. Yeah. So those are three different AIs: artificial intelligence, artificial insemination, avian influenza. I want you to rank those from best to worst AI. Oh, yeah. That's a good call. Okay. Um- number four on the list, 'cause there's another AI that's close
to me that I'll sprinkle in here. Number four on the list would be artificial insemination. That's kind of old news now. Important to me. My whole business is built on it. And- Yeah, yeah … I have Claude working on a whole thing looking at, semen density and things like that in our production. So does Chad, but it's unrelated to work. And then avian influenza. I'm thankful I don't have birds, and I feel really bad for people who do.
And then number two would be artificial intelligence, as we've been talking about. And number one by far is AI, AKA The Answer, Allen Iverson. Oh. The original AI. Man, great. Boom. For a gentleman of a certain age, that's a good one. Yeah. And that's part of why it's hard for me to see guys wearing short shorts, because I am from the Allen Iverson long shorts generation. Oh, yeah. Yeah. True, true. Yeah. They'll come back. They'll come back.
'Cause I have this theory that what people want to do is dress like their parents when their parents were cool. I think that's where fashion comes from. 'Cause your parents are cool before you're born, and then you make them uncool. Yeah. So your f- earliest ideas of cool are pictures of your parents before you were born. It's definitely cyclical, and I'm hoping when my kids are in college that they're gonna be wearing multiple popped collars. That is what- Yeah. Oh, that would be good … that's what I'm hoping for. Yeah. Man, that was a thing, oh.
Yeah. Come on. Okay. Chad and I, one of the things we both love are cool cars. And under appreciated cars, vehicles from yesteryear that we wish they were still being made with today's technology. Yes. Tell us about the Honda Element. Todd, this is, this is an open wound. This is an open wound for me. Oh. I had to give an official condemnation of two of our Honda Elements today.
Oh. Oh. Today. Bad day in the shop. Maybe you can report it later if you're still- So, no, I'll soldier through but talk about their greatness. Maybe that'll help you through it. Push through. I'll soldier through. A long time ago, we, well, my dad won a lease on a charity auction on a Honda Element in the early 2000s, and no one was willing to drive it, 'cause they were so ugly. So he drove it around as his, boss vehicle. It looks like an iPod. It looks like an iPod. Yes. Yeah, it's just a box on wheels, but it's all-wheel drive.
There's no carpet. The seats fold up, and it's a little Honda, so they go for forever. Well, we, rather than buying farm trucks, we need to transport people around a lot of times, and then we live along the lake shore, so we get a lot of snow, so all-wheel drive is important. So we end up starting to buy these things. Next thing you know, we got 15 or more of these in our fleet. They stopped making them in 2011, which was terrible, but we got a ton of these around because they're kind of like a little farm truck.
So you notice them as we drove around is that they were lined up in different spots. Yeah. Great vehicles, they get really good gas mileage. They're safe. That's our- They've got suicide doors, and you can open them up They get- Basically it's got- … suicide doors. Two suicide doors, yep. Yeah, so it's basically you can open- Yeah … up like it's like a hatchback, but from both sides. How many… Did you say how many Dykhuis Farms has owned? I think we have 15 on our asset list from now, but there's been a few that have died in deer accidents or rollovers so at least we've owned at least 20 of them.
Wow. There's no one that knows the used Honda Element market better than Dykhuis Farms. So that was my dream car. I, graduated college in 2003, which I think the Honda Element came out. That's the year they came out, yeah. Yeah. That's the year my dad got the- But it was, they were advertised- … orange one. They were advertised as the car for, cool, outdoorsy young men. And I was like, "I am the target audience," but I had bad credit. So I did not get one. And to this day, I've been like, "I should've… Has my Honda Element
window closed?" And Joe shows me that I still have that potential, provided I buy it just a little farther than Joe is willing to drive to get it. This is a- If you are willing to come back up and visit, Todd- Yeah I will promise you a Honda Element ride. Oh. Wow. So one of my best friends had better credit than me, that you could recline the driver's seat to be a bed It's also you can camp. They fold down flat. All the seats fold down flat, it's like a camper bed in there when you do it.
Yep. Such a cool- What an innovation. Now, you are the only person that's ever said that they were cool. I was thinking we were gonna talk about muscle cars or something. Nope. Nope. I feel like if everybody knows it's cool, it can't be cool. It's gotta be underappreciated to be truly cool. Yeah. Texas Roadhouse, great. Everybody knows it's cool. You gotta be in the know to know the Honda Element's cool, I think. Geeking out on economics is definitely cool then, Todd. You got me to share that trade secret.
And actually I was thinking about this on the… Do I talk about this stuff in public or do I just keep this a secret to myself? But it occurred-- So my, my son is 16, he's a junior. He's going to the same school that I went to and he is taking economics this year with the same teacher that I had. So shout out to Mr. Baker, at Holland Christian High School. The follow-up for Joe after day one of school was, "Where's that econ book?
I wanna take a look at it." Right? And they're like, "Sorry, we have an app on the computer." He has a book. He has a book for that. Way. Yep. Yes. My kids have no books … I'm sitting at my counter, maybe 11:00 at night. I've got Claude in front of me, and I'm flipping through this junior level econ book. I think the term is comparative advantage, right? It's a competitive advantage that even if people know about it, it doesn't matter. There you go. I think that AI and Honda Elements are a comparative advantage for Dykhuis Farms.
So it's okay to talk about them, and all I can do is gain some more knowledge that can help me further that comparative advantage. Take that PT Cruiser. I, know nothing about production of audio files, but that's the sound bite right there. That's just dead on. The Honda Element and the AI are the comparative advantage. I feel like you're just our target listener, Joe. We might be making this podcast for you When we started this- Mm-hmm … and we've talked about it on the podcast, but, the truth is they were
like, "We just need more content." And Chad and I are like, "We'll do it, but we wanna do whatever we wanna do." And they were like, "Well, what podcast would you wanna make?" And we were like, "Something we would listen to." So that was, literally just our only North Star. Like, we would wanna listen to it. And they're like, "Do other people wanna listen to it?" And I was like, "Does not matter. If the goal is for us to make stuff-" Content "… and we wanna do stuff we would listen to- and then you, Joe, have represented to us there are other people out there. Yes, there's- And they know about the Honda Element.
And you entered a market where supply and demand were not in balance, so you entered on the supply side with an advantage. So good for you. Okay, so you have a Honda Element that is a time machine, and you can only use it for work purposes. When are you going to with your Honda Element time machine? I think I need to go to the future. The thing that I'm hyper-obsessed with right now is disease risk, risk profiling, risk management. As well as consum- I'm perplexed by consumer demand right now.
I wanna go into the future and see what actually happens with pig diseases to know if these extravagant investments in filtration technology are something that the next generation is gonna say, "Wow, those guys were dumb." Or if it was really something that was needed even if the diseases we are targeting now had a cure. I'm pretty sure there will be disease in the future and maybe it will be
different, but I'd really love to know how that plays out, 'cause that has a lot to do with my financial projections. The other thing is will… i'm not on social media, so maybe if I got on social media I would get it more. But economics is all based on people making rational choices, right? And I wonder, maybe I wanna go back and meet with Adam Smith and learn about- how he came up with the initial, theories of economics. 'Cause I'm sure he would think that we were crazy now.
But they've kinda held up. So will they hold up? And maybe I just don't understand the rational choices that people are making. So I wanna go both. I wanna go forward, I wanna go backwards. But he's a good user of a time machine. You're a classic farmer. You wanna put hours on that time machine. Yeah. Absolutely. I'm depreciating this thing. I might as well use it. Absolutely. For sure. If it's a Honda time machine, it's gonna go 300,000 miles. So. Oh, that's true. Todd, this one that went dead today, 297,000 miles.
No way. Yeah. That's farther than we are from the moon. Yep. So, and it's- Did you have some words for it in the shop? Did you lay your hand on its hood I, no, I don't, I'm not sure. I don't even know which color it is. Yeah. So. I- You're like, "We're gonna bury it." I can't connect with it emotionally yet. Anyway, Joe, thank you for being so generous with your time yet again. Joe, you mentioned that you're close to the lake in Western Michigan. So we toyed around with this holding research retreats where we would go and work hard, but then have some place that we can recreate to think
through what we've talked about in the morning and then come back. Would you be amenable to us, using your conference room? Like in the off times, not in a peak time. We don't wanna be there, we don't wanna be there peak recreational season. We don't wanna be there in peak Dykhuis Farms time, right? How would you feel about us coming and bugging you for another day? You're welcome anytime. Thanks, Joe. Subscribe and download. Yeah, just, just come back.
Come back in a month. We'll keep learning about the ag economy together. How about that?
Buying more farmland is a common way to grow a farming operation. But with tight crop margins, high land values and borrowing costs, is it always the best path to profitability? On this AgCast episode, hosts Chad...
How do farmers make decisions - and what can the growing amount of data generated on farms tell us about those decisions? If you've followed Purdue's Crop Basis Tool or the Center's monthly basis updates, you've...
Farm equipment markets are constantly changing, but what can auction data tell us about where agriculture is headed? In this episode of the Purdue Commercial AgCast, hosts Chad Fiechter and Todd Kuethe are joined by...
AgCast is back - with new vigor but the same goal: bringing farmers and agricultural leaders together for conversations about the ideas, businesses, and decisions shaping the future of agriculture. In the first episode...
Margins are tighter. Interest costs are higher. And many farms are discovering that profitability and cash flow are not the same thing. In this episode of the Commercial AgCast, Chad Fiechter and Josh Strine sit down...