AI agents are making their way into agriculture, helping turn increasingly complex farm data into actionable insights. In this Feed & Grain Chat, Brendan Bachman, senior agronomy technology manager for GROWMARK, discussed how the cooperative is putting that technology to work through its AI Agronomy Agent within the myFS Agronomy platform.
The tool combines weather, planting, scouting and remote-sensing data to help crop specialists and growers evaluate crop conditions, identify disease pressure and prioritize field scouting. GROWMARK is also using AI to identify nutrient deficiencies and evaluate the potential ROI of in-season decisions. Bachman said myFS Agronomy logins and user activity this year are about 250% higher than the total recorded during all of last year.
He also previewed GROWMARK’s next AI agent that will ask follow-up questions to provide more context-specific guidance while keeping human expertise at the center of decision-making.
Transcription of interview with Brendan Bachman, senior agronomy technology manager, GROWMARK:
Elise Schafer, editor, Feed & Grain: Hi, everyone, and welcome to Feed & Grain Chat. I'm your host, Elise Schafer, editor of Feed & Grain. This edition of Feed & Grain Chat is brought to you by WATT Global Media and FeedandGrain.com. FeedandGrain.com is your source for the latest news, product and equipment information for the grain handling and feed manufacturing industries.
Today, I'm joined by Brendan Bachman, senior agronomy technology manager for GROWMARK. He's here to tell us how GROWMARK is better serving members across the coop system with its newly launched AI agronomy agent. Hi, Brendan, thanks for joining me today.
Brendan Bachman, senior agronomy technology manager, GROWMARK:
Yes, it's a pleasure being on, and it's been an exciting journey thus far. And we've got a long way to go, but looking forward to the direction of the future.
Schafer: Well, thank you for coming on and sharing it with us. So, we're a few months post-launch of GROWMARK's AI Agronomy Agent, so what results are you seeing in terms of time savings, crop specialist productivity or grower engagement?
Bachman: Yeah, that's a great question. When we launched this particular feature of the myFS Agronomy application, I would say it was probably at the highest anticipation of anything that we've had as a feature set inside of the tool thus far. And so, when we think about the excitement and use cases that we've seen thus far, it's been plentiful. And I think the reality is, the great part of AI is that it's iterative.
And so before, we had to take all this different feedback of a particular tool and a function, spend six months developing to try and get an outcome that met the user request. Well, with AI on top of all the data that we have inside of this hub and spoke model of myFS agronomy, the independent users are creating a whole host of different value add pieces at that local level.
Anything from, ‘How did the plants emerge and what are some of the yield-limiting factors?’ that they can just query from the top. Looking at planting dates, looking at crop staging and lists to help them with operational demands through what we're really looking forward to, [which] is some of the results once we get yield data in here on hybrid performance in soil and weather characterizations to really help us position products more effectively when we think about yield and environment characterizations. And so, the individual use cases are plentiful.
Right now we're getting a lot of really cool things that are being created around disease based upon the information we have in there. whether it's from scouting reports, boots on the ground, or some of our disease weather models. When you put AI on top of that and ask it a very pertinent question, like, ‘Where am I seeing disease?’ the reports that it's kicking out, giving our agronomists the ability to get into the fields, the right fields and right locations to validate, but also to have those really progressive conversations with growers on disease prevalence, as well as how the crop is performing overall, because that fungicide decision does need to have economics. And if I have fields that are performing well or underperforming, maybe based upon huge amounts of rainfall that we've had here in the last three weeks in central Illinois, we need to be taking those into consideration and that AI tool gets us there quick.
Schafer: Can you share some examples of how the AI agent was used to help a grower make a management or profitability decision?
Bachman: Yeah. The first one that comes to mind is, early on, it's kind of been the tale of two seasons, as far as we're sitting in like 105 heat index today, but we actually had some really cool weather early on. There was some frost risk that was present in Northern Illinois, Iowa, and I know one particular grower that just queried, ‘Where should I be looking on my farming operation for frost damage?’ And it sorted the fields based upon the relative weather information, based upon some of the topography. Frost lies low inside of fields and that's where the most damage occurred, and it took him to the fields and prioritized his scouting, if you will, to go out for a replant damage assessment on frost. That particular grower did end up replanting based upon that lead indication and where he should go look for those particular frost damage pieces.
And that was exciting to see because in a lot of those cases, what AI is going to be able to provide us with is good management practice and decision making capabilities. And when we can step beside a farmer and provide them information, that honestly has nothing to do with us positioning a product but just helping him make a better decision on his farming operation, like ‘Where should I look for frost damage and how can that impact my decisions to replant?’, I think is a big win for us, as well as the technology and obviously that grower.
So, that's first one that came top to mind. There's several in this mid-season where we've been able to identify nutrient deficiencies and quantify those quickly based upon remote sensing and bringing together ROI and economics to help growers assess, ‘Do I need more nitrogen based upon denitrification?’, or other types of pieces. And I know growers have been utilizing the tools that the agronomists are putting in front of them to help assess the ROI of some of these more urgent in-season decisions, as well.
Schafer: So in this time, what have you learned about adoption and how have crop specialists and growers responded to AI generated insights?
Bachman: The response has been huge. Currently, year-to-date, over all of last year, we're about 250% up in logins and user activity. I equate a significant portion of that back to the ease of use and the additional insight of bringing together information through the AI agent and the capabilities that are present. When you look at the myFS agronomy tool, there's a lot in there. And it's all specifically designed to do certain things, but growing a crop and monitoring it is very dynamic, and when you think about all the information that is present inside of a crop management tool, AI is really helping us sort through the chaff and get those insights pulled to the top and very easily understood in usable ways based upon a natural language question.
And so that connection of all the data to like relevancy and response is becoming a lot clearer through these types of applications, and I think that's driving the adoption curve. It's not to say that what AI says is the gospel. It does still require human elements, intervention and perspectives on what is that right decision. But — I'll use a crop specialist quote that was told to me in conversation — ‘It gets rid of 70% of the work that I don't like and allows me to focus on the 30% of my job that I really love,’ which is, solving problems for growers, not going through piles of data and information to try and figure out what's relevant. And so, I think we're seeing that adoption curve increase because it's doing just that.
It's getting us inside the red zone — using a football analogy — very quickly, and then that human agronomist, crop specialist, grower interaction comes around that data to make that final decision on what to do to remedy the potential opportunity or problem in front of us.
Schafer: Where do you see technology going next? Are there any new capabilities or enhancements planned for the agronomy AI agent or myFS?
Bachman: The pace is continuing to become exponential in what we can do with AI, not just the agent. I'll back up a second and just speak about AI inside of the development timeframes and curves of technology. Things that were taking months in development are taking weeks because an individual programmer could only have so much capacity, historically. But them utilizing AI alongside their jobs of programming features, functionality, doing validation checks, etc. is allowing them to have 2, 3, 5x amount of work get accomplished in in a week. And so when we think about what's next, the pace in which we're going to continue to see enhancements on tools of technology is something I think we're experiencing now in our consumer life, but specifically to this myFS agronomy tool, is what you're going to see in the coming months and years, as well.
So, that is a background talk very practically that we're already getting ready to launch phase two of our AI agent over the next week or so. And what phase two is going to bring in is source documents, logic that really takes what is all the information and context that the AI agent could be surfacing and bringing that down to think more like an agronomist than a technology tool. What are the pertinent questions that I should be asking the grower or the crop specialist, if you think about it that way, to uncover more information that is relevant to the decision?
And so today, AI agent 1.0, you ask it a question, it gives you a response back based upon the context of that question. In 2.0, we're building in the logic that if you were talking with a person, they might ask more questions: What herbicides have you put on already? What was the planting date? What are some of the other considerations that should be known to give you a better decision? And that's going to be interactive to where the agent is going to have the context of what questions to ask back to the individual user to help narrow up its logic and decision-making process to get you the closest to the right answer as possible. And the progression of that is happening in two, three months, as well as a whole host of other things that are coming inside the next version.
Where it's going in the future? I don't know. It's just moving so fast in ways that I think we have been excited for and anticipating, but now that it's here, the pace is something that has probably been surprising on how quick we can continue to iterate and drive additional value.
Schafer: Well, exciting things to come for sure! Thank you, Brendan, for sharing your insights with us today.
Bachman: Yeah, absolutely, I'm always happy to talk about it. It's exciting stuff, and I think we are in the throes of technology really taking hold and becoming that practical data insight that we've been talking about for the past 15, 20 years, so it's an exciting time in agriculture.
Schafer: Absolutely. Well, that's all for today's Feed & Grain Chat. If you'd like to see more videos like this, subscribe to our YouTube channel, sign up for the Industry Watch daily eNewsletter, or go to FeedandGrain.com and search for videos. Thank you again for watching and we hope to see you next time.