Jim Lenz, GEAPS (Host)
What if the biggest challenge facing your grain operation isn’t a lack of data? What if it’s the gap between having the information and actually knowing what to do with it? Everyday grain professionals make decisions that affect safety, quality, throughput, efficiency, and ultimately profitability. And those decisions are getting more complex. Facilities are generating more information, experienced employees are retiring, operations are under pressure to do more with less, and new technologies, including artificial intelligence, are changing what’s possible. But more technology doesn’t automatically lead to better decisions. So how do we close that gap? Welcome to the Whole Grain Podcast, where we connect you with the people, ideas, and innovations taping the grain handling and green processing industry. I’m your host, Jim Lenz, Director of Global Education and Training Institute, the Grain Elevator Processing Society. Today we’re exploring something that touches virtually every part of a green operation. Decision making. My guest is Shail Khiyara , CEO of Forum SWARM Engineering. Shail and his team are working at the intersection of artificial intelligence, optimization, and human decision making, helping organizations tackle complex operational challenges in entirely new ways. And our conversation isn’t really about technology for technology’s sake. It’s about the decisions people make every day. How do we take the enormous amount of information available to us and turn it into action? How can technology help capture and extend the knowledge of experienced operators? And what happens when artificial intelligence becomes less of a buzzword and more of a practical tool for solving real operational problems? That’s where we’re headed today. This is the decision gap how better decisions are transforming grain operations. We’re about to start my conversation with Shail Khiyara after this message.
Sponsor Message Strong America
Jim Lenz, GEAPS (Host)
When harvest is on the line, your dryer has to deliver. Strahl America brings more than five decades of grain drying engineering to operations like yours, manufactured in North America, are patented dictating mixed flow design drives evenly and generally, while unmanned heat recovery technology improves energy efficiency. So your brain comes out consistent batch after batch. No internal screen, no back work. People who take the time to understand your operation. Strahl America.com. Alright,
Meet Shale Chiera And Swarm
Jim Lenz, GEAPS (Host)
today we’re exploring how grain facilities can make better operational decisions using the data they already have while preserving the knowledge of experienced employees. Today, our special guest is Shail Khiyara. Shail, welcome to the show.
Speaker
James, thank you for having me here. It’s really exciting to be uh talking to GEAPS again and uh speaking with you today.
Jim Lenz, GEAPS (Host)
Yeah, we’re so excited to have you. Let’s tell our listeners a little bit about your background in the bus.
Speaker
Sure. Look, my background has been in the software and AI space over the last 20 plus years, but I didn’t start there. I started as a civil structural engineer and worked on a lot of material handling systems, uh, moving grain, moving uh seeds, moving coal. And that industrial knowledge, insight, and experience has really translated into uh what I’m doing with Swarm. And Swarm is a decision intelligence company that is uh focused on optimizing uh decisions that are made in supply chain and logistics and workforce planning, particularly at ag and ag food companies. So we serve the grain market pretty significantly grain milling, grain manufacturing. We are also focused on protein and produce, but grain is a significant component of our business.
Decision Latency On A Busy Shift
Jim Lenz, GEAPS (Host)
You earlier introduced me to the term decision latency. What does that mean?
Speaker
Yeah, it’s uh it’s essentially a decision gap. Picture this, for example, you have three trucks lined up and a bin is getting close to full, and somebody has to make a call, right? Split it, blend it, or hold it. And right now, that call rests entirely on one person’s experience. You know, years of knowing the facility, the grain, the time of the year, and knowing it well enough to make the right read in seconds, right? That’s not a weakness in the system. That’s the system’s greatest asset. The the challenge is the judgment is only in one person’s head, and it’s only as fast as that one person can make that that decision. So the time between an event occurs and you have the ability to gather data to make a decision, that’s what I call decision latency. It’s the gap of having what you need to know and being able to act on at the speed and the scale that the operation actually needs. It’s not that experienced people are slow, it’s that we’ve never given them anything to extend their reach to make the same good judgment available on every shift, every location, every moment, not just when the right person happens to be standing there, right? So used to be that the gap was forgiving, the decision latency was forgiving. Uh, you had a shift, a day to sort it out. And today, with the margins being as tight as they are, that same gap costs real money, you know, quietly every single day, not because anyone made a bad call, but because you know good judgment couldn’t be everywhere it needed to be at once. And that’s decision latency, the decision gap that you can actually compress by using AI augmented solutions by taking in your data, by optimizing, by giving you insights into your data so that you’re able to make a better decision and reduce that gap.
Jim Lenz, GEAPS (Host)
Great clarity, that makes sense. Can you give our listeners a few grain industry examples where delayed decisions quietly cost money? They do.
Speaker
And, you know, these are moments where the operator usually already knows to make the right call. They just don’t have the visibility or the time to act before the window closes. You know, take, for example, rail. You know, demarage costs in rail run between $75 and $300 or more per rail car per day, uh, depending on obviously the uh railworld and the equipment. And that sounds small until you scale it to a real facility. A shipper who is moving 50 rail cars a month, who averages you know, one extra day delay per car, is looking at around you know four thousand to eight thousand dollars a month. And that’s at the base rate before it escalates the longer the car sits. Um you run that across a full shipping season and across a facility moving hundreds of rail cars, and you’re talking about tens of thousands to six figures a year for delays that had nothing to do with the equipment failure. And the grain was there, the cars were there, the sequencing, scheduling call, you know, came uh a few hours too late. And that happens over and over again. Blending works the same way, right? Just a different mechanism, for example. If moisture or uh protein um isn’t caught until after uh you know the the load is already unloaded, you’re not managing the spec anymore. You’re correcting it after the fact. And uh discounting that load, running it back through the dryer, pulling premium grain to fix a problem, a cheaper blend could have solved. You know, all of this could be done if the call had made, you know, let’s say 20 minutes sooner. And that happens load by load, but a a mid-sized elevator is making the that call hundreds of times a season. You multiply that by a modest per load cost by that volume, and you’re really looking at a six-figure margin erosion. That’s money that never shows up as one write-off because it’s spread across every intake point, every shift all throughout the year. So these are some of the examples, and there there are then there’s one that isn’t about money, it’s about you know what you’re even allowed to sell. I was just at a Jeeps event two days ago in Omaha, and the topic of the day, besides AI, was mycotoxins, right? So corn under 20 parts per billion of aflatoxin is is fine for any use. Corn over 300 parts per billion has to be blended down with clean corn before it can even be fed. And that blending, by the way, has to be documented and permitted, and it’s not a quiet judgment call. So the decision on uh a hot load isn’t uh, you know, does this cost us uh a little? It’s do we catch this in time to segregate and blend it correctly? Or does it get co-mingled into a into a clean bin before uh the test result even comes back? And you know, you’re potentially compromised uh with everything else uh sitting in the bin with it. And that’s a decision that happens in minutes, particularly when the truck’s on the scale, not not after. That’s what makes decision latency or decision gap so dangerous because it costs you margin quietly all through the year. And sometimes it costs you the whole bin all at once.
The Quiet Costs Of Slow Calls
Jim Lenz, GEAPS (Host)
Yeah, but very thorough. This is uh making sense. Great examples. Are you kind of also stating maybe that these small delays often could be more expensive than a single equipment failure?
Speaker
Oh, absolutely. Because it’s look, the simple reason is the equipment failure tends to announce itself. The motor goes down, the leg jams, you know, everybody in the building knows about it in five minutes. It gets fixed, it gets logged. Someone puts a number on the downtime, the facility moves on, and it’s visible, it’s loud. I mean, I’ve been at at several of these facilities, and precisely because it’s loud, it gets managed. Decision gaps, or decision latency doesn’t announce itself. It’s like the difference between a leak in the roof and a dryer that’s running for two hours longer than it needs to. The leak you fix immediately. Water’s on the floor, it’s obvious, nobody debates whether it matters. But the dryer running long day after day, because nobody had real visibility into the into the moisture trend, nobody calls that an emergency. It’s not dripping on anyone’s head. It just quietly burns fuel and time every single day, all season. And by the end of the year, it’s cost more than the Roof League ever did. It just never felt urgent enough on any single day to fix. And that’s exactly the pattern that decision gaps, decision latency create across the board. It costs you a little constantly across every decision point all season, every season, and because no single instance is big enough to notice, nobody ever goes looking for it. So it just compounds quietly until at the end of the year the facility that seemed to be running fine left real money on on the table.
Jim Lenz, GEAPS (Host)
I think it requires pause to really examine your your systems, data that’s being collected. But this seems like, like you said, hidden or not known or not allowed. And so, but however, this is critically important. And you know, what also is critically important, and I have heard this a lot uh with my four plus years with Jeeps and personal conversations on the show of the whole grain podcast and a whole bunch of other places, is this grain industry is seeing so many experienced operators retire. Now,
Retirements And Lost Operational Judgment
Jim Lenz, GEAPS (Host)
what knowledge are organizations at risk of losing?
Speaker
Yeah, I’m glad you brought that up, James, because this is um this isn’t a future problem. It’s already happening, right? Workers 55 and older now make up about 23% of the U.S. workforce in manufacturing, which is as close a cousin to grain operations as there is, you know, nearly a quarter of the workforce was 55 or older, and almost 80% of the manufacturers said they’re seriously concerned about the exodus and what leaves with it. I feel grain is living through the exact same wave at the moment. And the thing that’s hardest to replace isn’t procedure, it’s judgment. You know, and it’s any facility can write down a checklist, you know, moisture targets, blend ratios, dryer settings. But what you cannot write down easily is the 30-year operator who walks past a bin and just knows something is off before any sensor flags it. Uh they they’ve seen this exact rain, this exact weather pattern, this exact time of the season enough number of times that their gut is basically running a model that nobody has ever coded. And that’s the knowledge at risk. Not the facts, but the pattern recognition built from thousands of small decisions over decades. And when that person retires, what walks out the door isn’t a manual, it’s every edge case that they ever quietly handled without anyone noticing. Because handling it quietly was the whole point. And the honest problem is that most facilities have no way to capture that. It’s not in a training document, it’s in the way the person watches the bin, in the way they read a truckload. Uh, and if nobody’s paying attention to that, the gap, the f uh, the the facility doesn’t find out what what is lost until the first bad season after that, the person’s gone. And you know, the same edge case shows up and nobody in the building has seen it before. So it is it is a critical uh gap in the market at the moment.
Jim Lenz, GEAPS (Host)
That’s that’s why I’m so happy to have you on the show. You recognize that. We certainly recognize that as an organization. I mean, we are redoing a lot of our online, on-demand courses, and one of them where we’re right in the middle of is a grain elevator equipment course. We have four SMEs that total 178 years of experience uh and looking at different scenarios and different responses to have uh the best learner-centric online courses we can uh make available to them. So it is it is so critical that we capture that and we’re grateful for that. And uh other courses that we have completed so far, we want to capture those experiences. Now, what’s so interesting at a time we’re at uh at this recording of the episode is that some people worry AI will replace people. You argue it captures and extends expertise. Explain that.
AI As A Decision Support Partner
Speaker
Look, I get why people worry about that, right? Anytime you put AI and um your job in the same sentence, the uh instinct is to brace for it. But think about what we just talked about, right? The the problem isn’t that we have too many good operators, it’s that once we don’t have enough of them, and the ones we we have can’t be everywhere at once. So what AI can actually do is take the 25, 30 years of operator judgment, the pattern recognition, the something’s off the with this bin instinct, and make a version of it available on every shift at every location, you know, not just when that one person happens to be standing there. It’s not replacing their judgment, it’s the closest thing that we have to cloning their attention span. Think about it like this, right? The operators built this instinct from the thousands of decisions over decades, right? Yes, indeed. AI can learn that same operational history, the scale tickets, the moisture readings, the outcomes, and start recognizing the same pattern that the operator did faster and hand that to that read to whoever’s up on the shift, uh, even the newest person on his or her first month. Right. The new hire isn’t as good as the veteran yet, but now they’re not starting from zero either. They’re starting with 30 years of pattern recognition sitting next to them instead of having to earn every one of those lessons the hard way, the same way the veteran did. And that’s not replacing a person, that’s making the expertise outlive the person who built it.
Jim Lenz, GEAPS (Host)
So AI can be seen as a decision supporter or decision support partner, right?
Speaker
Absolutely. Absolutely.
Jim Lenz, GEAPS (Host)
Could AI become a digital mentor for younger employees? Do you see that happening in the future?
Speaker
I do. I would go further than could it. Uh, I think it has to. Look, early in my career, I I structurally designed a material handling unit for a grain facility in Lake Charles, Louisiana. It’s still there, it’s still standing. It moves on tracks, so uh horizontal movement. It rotates the full 360 degrees horizontally, and and it uh swings zero to fifty degrees vertically, and it places material on both sides. On paper, that’s a math problem, right? Loads, stresses, tolerances, but the actual operating parameters, how it needed to move to match how grain flowed onto the site, that instinct from the operators who had run the equipment like this for years. They knew that the drawings didn’t capture, you know, where the belt would bind, where it would drift, what a normal vibration felt like, you know, versus a warning sign. And I learned early that engineering gets you to the machine. It’s the operator judgment that gets you the machine running right. And I didn’t learn that by designing and putting in blueprints and shipping it off. I learned it by being on the catwalks with the operators, standing where they stood, actually seeing how they ran it, not just how it was uh you know drawn to run. And that’s how mentorship uh used to work and still works in this industry as well. Uh a young person comes in, gets paired with someone who’s been doing this for 20, 30 years, and learns by standing next to them. You’re watching them making calls, asking why. And that’s how judgment gets passed down. It isn’t a manual, it’s proximity, right? Yeah. And the problem now is you don’t have enough of those veteran operators left to go around. And the ones you do have can only mentor, you know, one or two people at a time in one location in one shift. That doesn’t scale, especially with the retirement wave we have uh we have talked about. So what AI can do is sit next to that new hire the same way the veteran used to, not replacing the conversation, not at all replacing the relationship. But when that new operator is looking at a bin or a load and isn’t sure, you know, instead of guessing or waiting for someone more experienced to walk by, they have something that says, here’s what that pattern actually means. Here’s what’s happened in similar situations before. That’s not a textbook answer. That’s the same kind of pattern-based judgment that the veteran would give them. You know, just available in the moment that they actually need it. So I think over time the new operators are still building their own instinct. They’re still the ones making the calls, still the one learning, but they’re not learning from zero. And they’re not, you know, uh only as good as whoever happened to train them. They’re learning with 30 years of pattern recognition standing next to them, with AI augmenting that uh that expertise. And that’s what I mean by AI as a mentor. It doesn’t replace the the human relationship of mentoring, but it makes sure nobody’s flying blind while they’re waiting for one.
Jim Lenz, GEAPS (Host)
That makes great sense. I totally understand that.
The Hidden Data Already In Your Systems
Jim Lenz, GEAPS (Host)
In earlier conversations, you know, you believe that companies already have more data than they realize. What data were you talking about?
Speaker
Yeah, every every facility is sitting on more than they think. You know, take one truck, take a load uh coming in on an ordinary day, it gets weighed, that’s a scale ticket. With you know, moisture weight, time of day, which fields it came from. It gets probed for quality, you know, protein test weight, mycotoxins we talked about, check on that. And somebody decides which bin it goes to based on you know what’s in there and what it needs to blend uh eventually, right? And that bin might have a temperature cable or a moisture sensor on it. Uh then there’s weather data that’s sitting behind all of this, free weather data, public weather data, and it means something different depending on where you are. You know, for example, over the course of the last many months, I’ve been in Nebraska, I’ve been in Iowa, I’ve talked to grain facilities that are based in Dakota, and a weather spell in Iowa affects moisture and drying decisions completely different than the same weather pattern that would be, let’s say, in western Kansas, where you’re dealing with a drier crop and a completely different risk profile. Or, you know, North Dakota, where the harvest window and the storage decisions are shaped by an entirely different climate pattern. And that’s just not detail. That’s difference between the right call and the wrong one on the same data. So zoom out of this, and you’ve got maintenance logs on every piece of equipment that moved grain. You’ve got inventory history across every bin, not just where you know one particular load went, but the pattern of what’s moved, where, all season, every season. And none of that was collected for decision making. It’s really exhaust from running the business, right? The scale tickets exist because you need to know it for a transaction and for recording purposes. The moisture reading exists because you need to know if you can, you know, bin it or not, but multiply that across every load, every piece of equipment, every season. And it’s a genuinely enormous amount of information. Most of it is just sitting in separate systems, never talking to each other. So that’s the part I feel people miss. It’s not that the data doesn’t exist, it’s that nobody’s connected, you know, file cabinet to file cabinet, one system to the next, to see what it is actually telling you, and telling you differently, depending on you know what regional location you sit in.
Jim Lenz, GEAPS (Host)
I was gonna add to what you just said there at the end. If data already exists, why isn’t it producing better decisions?
Speaker
Because having the data and having the picture are two completely different things. You know, if we go back to the you know, one truck, the scale ticket lives in one system. The quality test might live in a lab spreadsheet. The bin sensor data is in whatever platform came with the with the different sensors, right? The weather is not even inside your building, it’s a website somebody checks separately if they check at all. Uh maintenance logs are often probably on paper. And none of those pieces is necessarily accurate, but none of them are are wrong. The problem is nobody’s standing at the intersection of all of them at the moment the decision needs to be made. And it’s the operator deciding where to bin that load, and they don’t have the the mechanics maintenance history in front of them as an example. Right. The person managing the dryer schedules doesn’t necessarily have the weather pattern sitting next to the moisture readings, for example. And that’s exactly where it leads to a bad decision, not a bad system. So that’s why more data hasn’t automatically meant better decisions. Disconnected data is functionality um uh the same as as no data. If anybody can pull all of it together in the 10 seconds they actually have to make the call, that would significantly narrow the decision latency or the decision gap and and stop margin from bleeding out of the system.
Jim Lenz, GEAPS (Host)
Yeah, you’re right there. Those dollars are critical, time is important. So the goal isn’t more data, it’s better decisions with that data.
Speaker
Exactly right. And I’d say that the single biggest misconception is that you know organizations don’t have enough data. Every facility I talk to, the instinct when something’s not working is we need more sensors, we need more data, we need more visibility. And sure, sometimes there’s a real gap in that, right? But most of the time, that’s not actually a constraint. The constraint is that the data they already have never gets pulled together in time for it to matter. And adding more data to a pile of disconnected data doesn’t necessarily fix that. It just makes the pile bigger. The goal is not, you know, to know more, it’s to decide better, faster with what you had in front of you. And AI can absolutely help with that where you don’t have to spend an enormous amount of time integrating various different systems. That’s the whole shift, right? From collecting information to connecting it right at the moment, somebody standing at a at a bin or a scale house to make that call.
Automation Versus Autonomy With IRAA
Jim Lenz, GEAPS (Host)
This is a great time to jump to a new segment. We know about automation. I mean, automation versus now autonomy. Many listeners understand automation. How is autonomy different?
Speaker
Uh, it is different. Um, but before I answer that, let’s let’s just connect the dots of where we’ve been, right? Judgment is the thing that actually runs these facilities. And we talked about the fact that judgment is walking out the door with retiring operators. And the data that could preserve and scale that judgment already exists, is just disconnected. And that’s exactly what automation and autonomy comes down to, to your question, right? Automation follows the rules, and autonomy is what it looks like when that disconnected data finally gets connected into something that can add on the same judgment that the veteran operator had. Automation is what most facilities already have, plenty of it. You know, PLC is running the floor. In a lot of cases, these systems haven’t been touched in 20, 30 years. Some of uh your operators are probably looking at green screen interfaces on control systems, older than half the people working the shift. A dryer shuts off at a set moisture target because a PLC is uh programmed to do exactly that. You know, a conveyor stops when the bin is full. It’s the same logic. So they don’t notice anything outside of that rule, right? The the automation systems that exist. If the situation is different than what the rule anticipated, automation just keeps following the rule, uh, you know, right off of the cliff if it has to. And autonomy is different because it’s making a judgment call, not executing a rule. It’s looking at the actual situation, it’s looking at multiple nonlinear pieces of data, the load, the weather, the bin, the equipment history, and it’s doing what an experienced operator would do, not what a fixed setting says that it should do. And I build a simple way of thinking about this from um, you know, if you if you think about the distance between autonomy and automation, uh, it’s not just one leap, it’s it’s a few stages. So first you have to actually see what’s happening, the data has to be visible in one place. Then you have to be able to react to what you’re seeing fast enough for it to matter. And then and only then you get to point where the system can actually decide and act on its own, you know, not just a flag, uh, you know, but there’s a fourth piece to this, which is, you know, people skip past way too fast. Every one of those decisions has to be traceable. So there has to be accountability into this. So I call these four stages, right, uh intelligent, responsive, autonomous, accountable. IRA, IRAA. But honestly, these letters matter less than the order. You can’t skip to autonomous. Most facilities are sitting somewhere between the first two stages, intelligent and responsive, but don’t have autonomy and accountability. That’s completely normal. It’s a ladder, uh, not just a light switch that you can change from automation to autonomy.
Jim Lenz, GEAPS (Host)
That’s a great framework. Thank you for sharing that. Yeah, it’s a ladder. You can’t skip for that to make sense because we are trying to remove those disconnected systems, trying to remove the lack of context, providing that context, removing the silos and autonomy is quite different from automation. And so that’s uh a place where I think people are really thinking about. And and now, if you could, can you draw out or walk us through a real green facility example?
Speaker
Yeah, let’s let’s put the framework to test on the floor instead of uh you know in the abstract. Take blending as one come one example uh in a grain operation, right? Where we’ve already talked about how much that one decision costs when it’s late. So following that framework, intelligence stage would would look like this, right? The facility can see in one place the moisture and protein on the load that just came in, what’s already the in the bin that it might go into, and the weather that has been affecting it all in one place, right? That’s just stage one. That’s just having it all visible instead of five different systems. Uh responsiveness is the next step, right? The system doesn’t just show you that picture, it reacts to something close to real time. The load comes in slightly off spec, let’s say, and instead of finding out after it’s already unloaded, it flags it while the truck’s still on the scale. Uh, this load doesn’t fit where it’s headed. Here’s uh what it does fit, right? So that’s the responsiveness from the system. The autonomous piece is where it stops just flagging and and deciding the system doesn’t necessarily wait for someone to read the flag and figure out what to do with it, it recommends. Um or in a more mature step, it can actually execute with human in the loop, right? The call itself.
Jim Lenz, GEAPS (Host)
Yeah.
Speaker
So that’s the autonomous piece. And the the accountable piece is what makes it trustworthy instead of just fast, right? Every one of those calls has to have an explainability, a reason attached to it. So the person can actually see what the reasons were, not just a black box saying, trust me, but here’s the moisture reading, right? Here’s what’s already in the bin, here’s why this is the right call. So the human in the loop is still critically important, but the accountability piece actually creates that trust and the explainability that is required by the operator. So that’s the whole ladder, you know, on one load in real time, from seeing it to reacting to it to deciding it, and to be able to explain after the fact.
Jim Lenz, GEAPS (Host)
That’s great. And
The Next Decade Of Grain Operations
Jim Lenz, GEAPS (Host)
what does the next decade look like?
Speaker
Honestly, most facilities are still early, somewhere between seeing the picture and reacting to it in real time. Few are anywhere near true autonomy with real accountability behind it. That’s not a knock on on anybody, that’s just where the industry is. And I’d rather say that plainly than pretend otherwise. This isn’t going to be about machines taking over the floor. That’s that’s not the vision. It’s about making sure the the 30 years a veteran operator spent reading the bin or a cash load before it goes wrong doesn’t retire with them. It’s about that judgment finally getting the reach that it all always deserved and on every shift at every every location. And I think it does something else too, something we haven’t talked about yet, James. The industry, I feel, has a real recruiting problem. You know, young people don’t always see grain operations as where the future is being built. But if the generation walks onto the floor and finds real technology, real judgment-driven systems, standing right next to the best operators instead of a job that looks the same like it did 40 years ago, right? That changes the pitch. You’re not asking someone to inherit a job that is disappearing. You’re welcome welcoming them to um come learn from 30 years of pattern recognition on day one, working with tools that respect what the veterans uh knew instead of uh replacing it. And that’s a much better story to tell a 22-year-old uh than uh to say, come do it the way we’ve always done it. Uh I think the facilities that win in this decade will be the ones where systems earned real trust, where the decision can be checked, explained, understood, not just executed. And that’s what turns a fast system into one people actually hand real decisions to. So that’s the future. I’m actually very excited about it. You know, not less judgment on the floor, more of it everywhere where it’s needed. And a reason, most importantly, a reason for the next generation to want to be a part of building it.
Jim Lenz, GEAPS (Host)
I love this conversation. There are people who are been listening to Hulk Green podcasts uh for a long time, but I think through meditating and keywords, I think people are going to find this episode because they’re searching for this. They want to know what is that next step? I love the differentiation between automation and autonomy. That’s a really important discussion here. We want sound decisions in a relatively efficient manner that are based off of pattern recognition. There may be these listeners right now who are going, I understand this. It makes sense to me.
One Practical Starting Step
Jim Lenz, GEAPS (Host)
It makes sense that we may be leading or going in this direction, but then they’re like, okay, what do I do? So if a grain company listening today, Sheil wants to start this journey, what’s one practical step that you suggest?
Speaker
Look, there’s a lot of uh FOMO in the market, right? A fear of missing out on AI and AI is moving so critically fast. What should we do? Where should we apply the technology? And my my recommendation is don’t start with technology. Start with a walk around your own facility and ask one honest question. Where are we losing time before you know uh knowing something and acting on, right? So you don’t need a consultant for that. Every every plan manager already knows the answer in their gut. It’s usually the same two or three decision points everyone complains about, but nobody never, you know, actually measured it. So pick one of those. Pick one of those business problems that you are seeing in your operation. Not the whole operation, not a facility-wide overhaul. One decision blending, storage allocation, dryer scheduling, you know, logistics, whatever is costing you the most right now, and go find out what data you already have sitting around it. You’ll probably be surprised. It’s usually not that you need new sensors or a big investment to start. It’s that the scale ticket, the moisture readings, the bin history, uh, for that one decision may never have been looked at together in one place. And applying the appropriate constraints, the business variables around it in an AI model that can actually help you define a particular decision. So connect that one decision and see what it what it tells you. And that sounds simple and conceptually it is, but doing it for real is is is harder than it looks, right? Great companies are not necessarily the AI uh uh data scientists or uh developers, right? That’s not the core business. But having this data, working with organizations that can build an AI model specifically for your business, that is vertically attuned to your business, that understands the domain, that understands the ontology or the language that you know your particular uh business speaks, that is critically important. It’s not the horizontal AI solutions that you see out there, and there are many and they’re beneficial. Um, you know, Claude and Gemini and Copilot and others. Uh it’s really the vertical specific AI models that understand your business. Uh, those are those are critical. And I think if you pick a particular area, if you pick a particular decision that you want to uh automate, um, that is a great starting point uh for you to have. Uh you just have to prove to yourself, you know, on one decision, that connected data tells you something that disconnected data never could. And once you feel that, the rest of the path makes a lot of sense.
Jim Lenz, GEAPS (Host)
Great actionable advice. Great actionable advice and very practical. This is something they can all do.
Why This Future Is Exciting
Jim Lenz, GEAPS (Host)
Uh finally, Shea, what excites you most about the future of operational decision making?
Speaker
Look, we’ve we’ve talked about the cost of slow decisions, the judgment walking out the door, the data sitting disconnected, and what it looks like to finally connect it. But you know, if I’m honest, what excites me most isn’t any single piece of that is what this moment actually says about an industry that doesn’t always get called cutting edge. You know, grain may be thought of as old school, you know, trucks, bins, dust, decades old equipment. But I think this industry is about to become one of the more interesting places in America to build a career precisely because it is sitting on so much unrealized advantage. You have you know real physical operations, real consequences, real stakes, food safety. Um and that’s exactly the kind of environment where good judgment matters more than anywhere else. And uh where getting it right actually means something. So for me personally, having spent many years in industrial operations before I ever thought of AI, you know, I worked on gas oil separation units, uh, built from scratch, standing on you know, catwalks with operators who knew more than any drawing could tell me. What excites me is that I get to spend the next chapter making sure that the kind of hard-won knowledge doesn’t disappear one retirement at a time. That’s not abstract for me. That’s watching something I I respect deeply finally get protected instead of lost. And if you want the picture of where this goes, you know, walk into a grain facility 10 years from now. And the person on the shift, whether they’ve been there 30 years or 30 days, has the same read on that bin. Not because everyone became an expert overnight, but because the expertise finally learned how to stay. And that’s the future I’m excited about. Not a facility with fewer people, but a facility where nobody’s ever flying blind again.
Jim Lenz, GEAPS (Host)
Great
Final Takeaways And Next Steps
Jim Lenz, GEAPS (Host)
support of decision making ahead here for this. If people want to reach you, your CEO of Swarm, if they want to find out more information, House uh, or what is the best way to connect with you and your organization?
Speaker
There are two ways. Uh, one is visit our website, which is www.swarm.engineering.com, or you can reach me on email as well, uh shale at swarm.engineering. And I’m also very active on LinkedIn as well.
Jim Lenz, GEAPS (Host)
Fantastic. We’ll put that in the show notes. Make sure that people have quick access to that to go to those resources. And uh, Shail, it just means a lot that you are here. This is obviously a discussion and and there is some uh misunderstandings out there. I like that you, you know, one of the first actionable things is kind of going analog, uh, you know, uh, not quick AI or digital, is to really examine where are those things. Like you said, a lot of people may know where that is, but take a pause. What is one area or two areas that you could focus on? Start from there. And we know that expertise is leaving, and now is uh the time to really look at not only just automation, but autonomy. It’s about good decision making, like you said in the IRAA framework you provided there. This means a lot to us at GEAPS that we have great individuals and organizations who are there who are supporting the industry to help feed and fuel the world, keep the world safe, but it also starts with better decision making. And so this episode is a great resource for people, and uh that’s why I’m so happy that you are here as a guest on the show, and I want to thank you for your time and swarm uh as well.
Speaker
Thank you as well, James. James is doing a great job in supporting the industry, and I really love what you’re doing with these podcasts, James. Thank you.
Jim Lenz, GEAPS (Host)
Thank you. Good day. There is a phrase from this conversation that I think is worth holding on to the decision gap. Because having more data isn’t necessarily the same thing as having more clarity, and having more technology doesn’t automatically mean we’re making better decisions. The opportunity is what happens when we bring those things together. Operational experience, good information, human judgment, and increasingly powerful tools that can help us evaluate possibilities that would otherwise be difficult to see. For brain operations that could have implications for everything from throughput and efficiency to quality, maintenance, safety, and the way knowledge is carried from one generation of employees to the next. And perhaps the most important question isn’t simply what can artificial intelligence do? It’s what decisions could we make better? That’s a very different place to start. JL, thank you for joining me and sharing your expertise and perspective with the whole grain podcast audience and to everyone listening. Thank you for spending some time with us. Whole Grain Podcast is here for you. It’s brought to you by the Grain Elevator Processing Society. The International Professional Association is dedicated to advancing the grain handling and processing industry through knowledge, connection, and professional development. And we want to thank you, listeners of the Whole Grain Show, the Whole Grain Podcast has listeners from 113 different countries. This is a great educational network where we can take the learning on the goal and continue listening. Share this episode with a colleague and friend at work. Check out our website, Jeeps.com, that’s G-E-A-P-S.com for more information on how you can become a member, explore our educational resources and solutions, and in-person events such as GEAPS Exchange. I’m Jim Lenz, Director of Global Education and Training at GEAPS, where we work to reach the next generation and the current generation, the workforce in the grain industry. Until next time, keep learning, keep asking questions, and keep moving the grain industry forward.
