[🔊 Ep. 16] Farming with Data: How Advanced Analytics are Transforming the Agriculture Industry with Dinu Ajikutra

Experiencing Data with Brian O'Neill (Designing for Analytics)
Experiencing Data with Brian T. O'Neill
[🔊 Ep. 16] Farming with Data: How Advanced Analytics are Transforming the Agriculture Industry with Dinu Ajikutra

Today we are joined by Dinu Ajikutira, VP of Product at CiBO Technologies. CiBO Technologies was founded in 2015. It was created to provide an objective, scientifically-driven insights in support of farmland economics. Dinu is currently leading an effort to productize what I found to be some very impressive analytically-driven simulation capabilities to help farmers and agronomists. Specifically, CiBO’s goal is to provide a software service that uses mapping and other data to predictively model a piece of land’s agricultural value –before crops are ever planted. In order to build a product that truly meets his customer needs, Dinu goes the extra mile–in one case, 1000 miles– to literally meet his customers in the field to understand their pain points.

On this episode, Dinu and I discuss how CiBO will help reduce farmers’ risk, optimize crop yields, and the challenges of the agriculture industry from a data standpoint. We also discussed:

  • Farmers’ interactions with data analytics products and how to improve their trust with those products
  • Where CiBO’s software can be used and who would benefit from it
  • Dinu’s “ride-along” experience visiting farmers and agronomists in the midwest to better understand customer needs and interactions with the tool
  • What Dinu has learned about farmers’ comfort using technology
  • The importance of understanding seasonality
  • The challenges of designing the tool for the various users and building user interfaces based on user needs
  • The biggest product challenges in the ag tech field and how CiBO handles those challenges

Resources and Links:


CiBO Technologies

Experiencing Data Podcast

Quotes from Today’s Episode

“CiBO was built on a mission of enabling sustainable agriculture, and we built this software platform that brings weather, soil, topography, and agronomic practices in combination with simulation to actually digitally grow the plant, and that allows us to explain to the users why something occurs, what if something different had happened, and predict the outcomes of how plants will perform in different environments.” — Dinu Ajikutira

“The maturity of the agricultural industry [with regards] to technology is in its early stages, and it's at a time when there is a lot of noise around AI,machine learning and data analytics. That makes it very complicated, because you don't know if the technology really does what it claims to do, and there is a community of potential users that are not used to using a high-tech technology to solve their problems.” — DInu Ajikutira

“In agriculture, the data is very sparse, but with our software we don't need all the data. We can supplement data that is missing, using our simulation tools, and be able to predict weather outcomes that you have not experienced in the past.” — Dinu Ajikutira

“To add clarity, you need to add information sometimes, and the issue isn't always the quantity of the information; it's how it's designed.I've seen this repeatedly where there are times if you properly add information and design it well, you actually bring a lot more insight.” - Brian O’Neill

“Sometimes the solution is going to be to add information, and if you're feeling like you have a clutter problem, if your customers are complaining about too much information, or that's a symptom usually that the design is wrong. It's not necessarily that that data has no value. It may be the wrong data.” — Brian O’Neill


Brian:  My guest today is Dinu Ajikutira, who's from CiBo Technologies, C-I-B-O, and he is a senior director of product management and marketing, and he's going to be talking to us today about ag tech, agricultural technology, and specifically, their platform that's a simulation and modeling tool. It's a decision support tool to help people working in the farming space understand the crops and the yields that different pieces of land can produce, and this is really fascinating. There is a lot of different stakeholders involved in the ag tech space, whether on the insurance side, the banking side, or actually farming products, CPG, or the actual farmers themselves, and so Dinu's going to talk a little bit about his process for going about productizing these analytics and the data that they have in this space to help people working in farming.

Brian: So, I hope you enjoy my chat with Dinu.

Brian:  So, welcome back to Experiencing Data. Today, we're going to be talking with Dinu Ajikutira, who is currently at CiBO Technologies, and this is we're going to talk about farming and plants and agronomy and ag tech. Is that what it's called these days, as I know it? Ag tech?

Dinu:  Yeah, ag tech is a big buzzword these days, for sure.

Brian:  Yeah. I'm totally stoked on this because I'm actually uh I live here in Cambridge, Mass, and I've got a vegetable garden in my background. "My background." In my backyard. I definitely don't have any farming in my background that I know of, but I actually love growing vegetables and all this kind of stuff, and I was like, "Oh, this is neat. There's data analytics. We got AI, and we have vegetables and food," and so this is some good stuff, and it sounds like you guys are doing some cool things at CiBO with simulation.

Brian:  Can you tell us a bit about what CiBO Tech does and what its software products are doing to help in the ag tech space?

Dinu:  All right. So, first of all, CiBO stands for "food" in Italian, and the company was built on a mission of enabling sustainable agriculture, and we built this software platform, the technology platform that brings weather, soil, topography, and agronomic practices in combination with simulation to actually digitally grow the plant, and that allows us to explain to the users why something occurs, what if something different had happened, and predict the outcomes of how plants will perform in different environments.

Brian:  Got it. And your role there, you're in product management, correct? You're a senior director of product management, so you're working on the actual interfaces and data products that are used by the people that purchase the software, I assume. Correct?

Dinu:  Yeah, that's right. I run the product management group. So, you know I create software products that meet market needs, or what the market needs. Yeah.

Brian:  Got it. So, can you tell me about who purchases the software, and is that the same person that will actually log into, I guess, is it a web-based tool or something like that, I assume, like a SaaS?

Dinu:  Yeah, it is. It's a SaaS. It's a web-based tool. Once we authenticate a user, they could log in and test it out. We are at a beta phase right now, so we are not commercial yet, and typically you know like, with the buyers and the users are different personas, right? The decision makers are managing PNLs, but the users, in the case of CiBO, are the last-mile providers for farmers. So these could be agronomists or CPG companies, potentially, or anybody who's actually serving the farmer.

Brian:  Got it. So can you walk us through, like if I was in CPG or if I was an agronomist, what is my problem that makes we wake up one day and say, "I need to get a demo of CiBO Technologies' stuff"? What is the problem there, and what is it that I'm going to want to sit down and do the next morning or whenever my subscription is enabled?

Brian:  Tell me about some of those use cases that you might go through.

Dinu:  Okay. Let me try, because there are various groups of industries, right, that are helping farmers do better, and the bottom line is, helping farmers making more money in a sustainable way. So the agronomists, there are retailers, retailing seeds or fertilizers. There are insurance companies that are insuring crops. There are bankers who are loaning to purchase land, rent land, or operate land, and so there are different groups of individuals that are interacting with farmers, with the expectation of improving the outcome for the farmer.

Dinu:  So, let's take an example of an agronomist. Agronomists are typically making recommendations of farming practices to get the best outcome from that land, and the biggest variable that happens in any field is the weather. Next year's weather is not going to be, at least, likely not going to be something we've observed in any of the years in the last five or 10 years, so that's the kind of challenge is, what do you expect, and how do you respond to it in a meaningful way to bring back the productivity back on track?

Dinu:  And that's a difficult deal. They're saying that you have 40 chances in farming, and that means to say that a farmer has 40 years of farming, and he learns every year, and every year, that's the only chance he gets to get productivity, but with technology, CiBO can help them make it 40,000. Try your hypothesis of what you want to do before you actually do it on the farm. There's a reason why we use dummies to do crash tests in cars, and it's something similar to that. You don't want to take the risk in the real thing.

Dinu:  So agronomists, the pain point for them is whether their recommendations are going to be the right one that actually produces the results or not. Using technology can prove it out before they recommend to the farmers.

Dinu:  These are the kind of things that people typically worry about.

Brian:  Got it. And so, are these agronomists working as like consultants to the farmers, and it's like they're like a service organization, and so this is a tool that they would use to improve the recommendations they provide to farmers, or can you talk about that a little bit?

Dinu:  Yeah. First thing there, Brian, is you know when it comes to farming, there is no standard practice. That's the challenge of agriculture. There is no certainty. In the past years that I worked on simulation and modeling for other industries like process industry, pulp and paper, airlines, building, it's all physical systems where you control everything. In agriculture, environment is something we don't control, and farming practices is done by the gut.

Dinu:  The first thing I want to correct right off there is, there is no standard practice, and if you can repeat the question back, I kind of lost track of what you precisely asked, and I can get back to answering.

Brian:  Sure. Yeah, so I guess what I was thinking about, so it sounds like, let's say in the pre-CiBO way, in the pre-being able to do modeling or running the simulations, you had scientists, these agronomists, that provided advice or recommendations to farmers on how to improve crop yield. That's basically what this comes down to, correct? What plants, maybe how to plant them or where to plant them, what's going to yield the best based on whatever inputs and information they had, and then the farmer decides how much they want to trust that or use their gut, and so your tool actually puts some science against this. It actually runs some of those simulations and allows them to provide better insights.

Brian:  So my question was, is it the farmer who's... I realize the farmer is probably eventually going to benefit the most, and then, of course, the insurance company or whoever, all the other second-order factors that happen afterwards, but is the farmer sitting down and actually using this software, or is it the agronomist, and the if it's the agronomist, do they act as like consultants to farmers? I don't know how that works, and they look at this as a better tool in their tool box for their consulting work, or how does that... ? Who would actually sit down and use the CiBO tech directly?

Dinu:  Right. Okay. All right. So, with an example, okay? So I'll give you a couple of examples. Let's say agronomists are typically in independent agronomist organizations. They belong to some groups that consult with farmers. Large farmers hire their own agronomists that do look after large farmlands that they own. There are also agronomists in large companies that provide agronomic services. These could be companies that are positioning themselves to holistically serve the farmers in all the different aspects that they need to deal with, and our technologies will be a tool kit in ensuring that they make better decisions in their recommendations.

Dinu:  So, let's say, an agronomic services company develops business plans for farmers. This will be one of the insights that will be inserted into that business plan on how they need to operate their farms as they respond to different weather events, et cetera.

Dinu:  Another possible user or users would be in the insurance companies or the bankers. In the case of the insurance, they are possibly looking at, how do you insure a particular level of crop outcomes from the land? Is it 190 bushels per acre is what they're willing to insure at? Will they give a premium for 200 bushels per acre? That's the kind of decisions they are making at the start of the planting season, which is around now.

Dinu:  The bankers, on the other hand, are also looking at the risk profile of the farmland to say, "Do I loan? How much money should I loan to this farmer, based on the outcome they want to expect from that piece of land from standard practices that the farmer has been executing on?" There are risks associated with each of these decisions. So the insurance agent or the banker might be looking at, what is the risk tolerance for that particular piece of land they're willing to accept to give a particular rate to the farmer? And our technology can help them determine that, based on our prediction tools.

Brian:  It sounds potentially from a user experience standpoint, it sounds particularly a little bit challenging. Do you serve up the same experience and the same tool set for someone that has a financial interest? I mean, I understand everyone has a financial interest, but I would imagine that, as you were talking about calculating a risk profile or something like that, the final information and the evidence you might provide to support the conclusions the software makes, I could see that experience differing for what a farmer wants to see versus what the insurer wants to see, right?

Brian:  They're looking for safety in their investment. "What is my risk of losing of this farm not performing well?" But I imagine they don't care as much about, "Which version of corn, you know which seed, whatever do we use?" I mean, I know that stuff's constantly changing, but they probably don't care quite as much about that detail the same way. Do they? I mean, I don't... Can you talk a little bit about, are they different products, or is it one product and you have to design carefully around all these different use cases that are here?

Dinu:  Well, first of all, let me qualify all my responses. We are at a beta stage, right? So, our users are still testing, and some of the things I'm saying, there are no real users who are using it today to make decisions, our product and technologies are capable of making these decisions.

Brian:  All right, so you're still productizing it, kind of, at this point?

Dinu:  Right.

Brian:  Ah, got it. Got it.

Dinu:  We are a startup. We've been in stealth mode for a while. We've built the technology. We can prove it to anyone who wants to see it.

Brian:  I see. I see.

Dinu:  And we have business going on. We have paying customers, but largely, much of the customer base that I talk about are still in the beta phase.

Brian:  Got it. Got it. Are those If you can't say, I totally understand; are those

more on the finance side or the farming side or the science side, and how do you go about testing that, making a good product a good data product out of this technology that you have?

Dinu:  We have to test. So, the challenge with agriculture... Well, there are two challenges in what I've observed. One is, this area's incredibly interesting, and that is a problem. It's interesting because you get digressed, and as a product leader or a product manager, the primary skill you bring to the table is focus. You need to understand precisely what you want to get done. You want to always communicate in writing what you are going to get done and ensure that you're getting that done in a timely way, and that is particularly challenging in that ag tech.

Dinu:  Secondly, the maturity of ag industry to technology is in its early stage, and it's at an early stage at a time when there is a lot of noise around AI and machine learning and data analytics, and that makes it very complicated because you don't know what technology really does what they claim to do, and it's a community of potential users that are not used to using a high-tech technology to solving their problems.

Brian:  Sure.

Dinu:  If you look at, so, now, remind me again, Brian, you were asking me about specific users and how they use it, and how do we determine that it's usable, correct?

Brian:  Yeah, I was curious. I guess I didn't understand initially where you guys were in your product lifecycle. So it sounds like you're still kind of figuring out what the product is based on these different problems sets that you know exist in the market, so how do you go about testing your ideas? Are you doing formal like experience testing? Or is it just conversations? How will you validate that the product itself is on the right track for these different constituents you have? Do you have a process you follow?

Dinu:  Yes, we do. We have a large base of scientists in the company. If you think about our target industries, right, there are going to optimizing farmland and farmland purchase and rentals. There is a part of optimizing operations of the farm and then optimizing the supply chain around the farm. Those are the three pieces.

Dinu:  Now, we have, in the company, crop scientists and agronomists that act as internal users for the agronomist use case or the scientist use case in terms of ag import companies, et cetera. We also have tie-ups with the large farm groups around the country who help us by testing our software and informing us about some of their pain points and use cases.

Dinu:  We conduct alpha tests and beta tests with literally field agronomists going to farmers, face to face, and getting them to use our software and test it out, not because we want to necessarily sell the software to the farmers, but any company that is in the last mile to serve the farmer is going to come up with the answers that the farmer needs to respond to, so the best way we can do is to get the product and its solutions qualified by the farmer, which makes it most meaningful for the constituents that we want to get it to.

Dinu:  So we have a whole bunch of different set of activities and programs that are designed to get different potential users to test out our software.

Brian:  Got it. Can you tell me a little bit about what that experience is like, both

for the farmer, but also for your team, for example, this agronomist? I'm picturing a guy driving up in a truck with a tablet, and he's like, "Do you have Wi-Fi?"

Brian:  "Oh, yeah, we have Wi-Fi."

Brian:  And he turns it on, and he loads up your screen, and he sits down, and they're having coffee at a table somewhere in Kansas, and you guys are showing him like, "Well, we ran this with you know the corn kernel A you know for next season, and we got 86% whatever, and then we did B, and check this out. Like we're seeing 98% blah, blah, blah. Here's what went into this. What do you think?"

Brian:  I'm totally making this up. Can you tell me about what that experience is? Especially because you're talking about this new tech in a space where people aren't used to using data to make decisions here. Walk me through that experience. Help the listener picture what this experience is like and how your product might fit into that.

Dinu:  Sure. In fact, whatever you said didn't sound like a story you are making up. It can be quite real. Actually, we underestimate the tech use by the farmer. The average farmer in the United States, in my opinion, having done no studies on it, by the way, use a lot more technology than an average middle-aged person, in my opinion.

Dinu:  Now, it's a crowded market. There's a lot of technology companies vying for the farmers' business. There's a lot of apps that the farmers could access, and they use a lot of technologies to make their decisions. Besides that, your question was specifically about, "Okay, how do you get around to actually interacting with this farmer, and how does that look in a typical day?"

Brian:  Yeah. I'm really curious how that feedback gets back to you or to your team to be properly prioritized. Are you guys sending any CiBO employees out to observe these interactions right from the horse's mouth, or is it mostly kind of like secondhand? Like the agronomist would report what he heard from the farmer, and then you guys interpret that?

Brian:  Can you talk about that?

Dinu:  Sure. Sure. So it's all of the three, by the way. We have agronomists, CiBO employees that actually go out to farmers. We hear from different farmer communities having looked at our technologies, by email or by conversations over the phone, but we, that is, product managers, literally go to farmers and sit with them and work through the product to ensure that it's something that they believe is valuable... and the part that I wanted to touch on here is my own visit.

Dinu:  I visited about five farmers or four farmers in Kansas and Nebraska about four weeks ago, and this was our field agronomist taking us to different farmers across 1,000-mile journey in his truck, and it's long roads. I've not been to that part of the country before, having not worked with the farming operations in the past. It's an exciting journey, and I'll give you one of the stories that I heard.

Dinu:  We went to this large farm. It's a large farmer, very successful farmer, and his strategy for growth is to expand his farmland and the farmland that he's able to operate because that's where his value-add is. He's very efficient at what he does. He's constantly looking for new farmland. The previous one that he had just rented, and he told us this, is about 30 miles north of where he currently farms and majority of his fields are, and he felt that he could've done little bit better due diligence before renting that farmland because it's not quite panning out like the way he'd expected, and he suspects it's probably the soil that's not doing it for him, and he's not getting the kind of productivity that he was expecting when he rented, and recently, he had a call from another farmer who wants to rent out his piece of land, so he was checking out our technology to see if this is something that would be meaningful for him.

Dinu:  The first thing he does is he wants to get comfortable with, "Okay, what can you tell me about my field? Because I know my field, and if you can tell me what I already know, I can believe that you can tell me the same thing about a farm that I don't know about."

Brian:  Ahh.

Dinu:  Okay, so he checks out his farm, and he looks at how his field looks in our solution, in our technology platform, and how the inter-field looks, how the stability looks. Do we identify the right zones as being the highest productivity zones, the low productivity zones, the unstable zones, and I can explain if you want to get into it what these mean.

Dinu:  Then he finds out how we predict how his field has performed, having known nothing about this field directly from him, but purely by the data we infer and our simulation models, we can tell him how he did in the last 10 years, and having convinced himself, now he is looking into the farmland that he wants to rent and understanding the same dynamics, and because ours is a simulation model, now he can change some parameters to say, "Look, I don't plant at that density. I usually plant at a different density," or, "I plant a different hybrid. How would that normally perform? If I invested in irrigation, how would that perform?"

Dinu:  And these are the questions he typically wants to answer, and we help him do that by just observing him and saying, "Okay, you can do this with our software. How would you do it, and is it intuitive enough for you to get your answers?" And by pure observation and recording, we can come back and get an improvement and then perhaps go back to him and say, "Can you test this out for us if you don't mind?" And we need to time it really well with the farmers, by the way, because probably, we have another week or so. Once they start planting, they don't have the time for this kind of stuff, so you got to get them in the right time, when they have the time to pay attention to you, and once the farming operations begin, a farmer gets really, really busy, and he can't pay attention to anything else.

Brian:  There's so much here. My brain's like on fire thinking about how this relates to so many other analytics and data products.

Brian:  So, one thing you talked about here is, there's clear seasonality here, literally the word "season", right, when we talk about farming, but so much of the time, I feel like, you know data products are not always designed with a sense that, oh, they're just always on, right? A web tool is just log in at any time, and it's there, but people aren't always thinking about, well, when is the right time to provide XY data to someone, some piece of insight? What are they doing in June that's different than they're doing in September?

Brian:  Like obviously, a farmer, when you're in harvest mode, you're in a very different mode than you are in planting mode, and so, just being aware of that, and you can't be aware of that if you don't get out there and talk to your customers and start to learn about what their life is like. This gets to empathy, so I love that you guys are literally doing a ride along, which, in design world, we call these "ride alongs" sometimes, and you're literally in a truck and visiting these farmers to understand this.

Brian:  So, the other thing, too, was the demo you talked about. You could see how this could become, "If your tool can explain to me what happened in the last 10 years, I'll start to believe the prediction." I think that's really powerful as well. It helps build trust that the technology, it's not magic, that it's actually looking at data points, and it's coming up with you know predictions I can believe and I'm willing to put my trust a little bit more in the software versus my experience, and I think that can be hard, right?

Brian:  I'm sure your tool will probably come up with some surprising recommendations at times that might really push the envelope, right? Like, "Really? You want me to plant eggplant," or, "You really want me to separate my crops to two-foot spacing instead of one with this variety?"

Brian:  Do you have any stories where the tool might've come up with a prediction that was a little bit unusual and you hit that friction, and your prospect or the participant you were talking to was a little bit, "I'll believe it when I see it," or any anecdotes like that?

Dinu:  No. I don't have any anecdotes to that effect, but the couple of things, the couple of points I wanted to make. One is that the farmer knows his farm more than anyone else does. They've farmed in that piece of land, and they've tested it. There's a lot of generational knowledge that's gone into it, but here's what we can do: we can tell the farmer about a farm that they don't have knowledge about. They have not put the boots on the ground on that particular piece of farm. That's where we come in and bring value, and we do it also in situations that the farmer has not encountered before, even in their own farm. There are situations they've not encountered in the past, and we can give them information about that, and that is a value of using technologies in their own farms.

Dinu:  The thing about data, Brian, and how we think about it differently is that data is you know is you can make a lot of sense out of data, but there are two critical aspects to that. One is, you need to have all the data that you need, and agriculture is not exactly an area where all the sensors are where you need them to be. So there a lot of the data is very sparse, there's a lot of gaps, and there's just not enough data for you to make the conclusions you need to make, and secondly, the past will not look like the future because of where we are in terms of climate change and the things that we're experiencing, so you cannot extrapolate biological systems from past observations to a completely different weather or environmental situations, and that is where we can come in and say, "We don't need all the data. We can supplement data that is missing, using our simulation tools, and be able to predict to weather outcomes that you have not experienced in the past."

Brian:  Can you tell me a bit about the experience? Like so let's say I'm that farmer. I think he was in Nebraska, right? And you're saying that he bought some land that he hadn't used before, and it didn't quite perform as well as he had hoped, or maybe he's looking for a new piece of land, and maybe he's trying to decide, "Should I go ahead and rent or purchase this land?"

Brian:  What is that experience you're going for? Like Is it load up Google Maps, put it in geo mode, get a picture of this area, and then I go into your tool, and I run some modeling? And like is this something that happens in real time? Is this a process that happens over many weeks? What's that experience like from the time when the farmer's like, "I really think I should go in on this spot 30 miles north of home, but I want some reinforcement before I pull the trigger on it"? Can you help us picture what that experience might be like, what you're going for from their experience perspective?

Dinu:  Sure, sure. From the user experience perspective, right?

Brian:  Yeah. Yeah.

Dinu:  Yeah. Yeah. They just come into our tool. There's an authentication system, either a username/password or they create on Google Authentication and get into the tool. When they come into the tool, there is a map. The map is your central navigational tool.

Dinu:  Now, there are several ways the farmer identifies his field or her field, by just typing in the address or uploading what's called a shape file, which is the coordinates of where the field is located, latitude/longitude coordinates, and typically, the farmer has shape files from different tools they have been using, so they can just upload it into our system, and immediately, when they do that, and it's a matter of seconds, although sometimes, the exportation of the shape files can take them some time, but if they have the address, it's the easiest way, they just get to the field, and after that, real time, we can show them their stability inside their fields in terms of their elevations and the depressions and the high-product-yield areas and the low-product-yield areas, et cetera, and you click on "simulation".

Dinu:  We already have all the inferred data in the planting densities, the planting seeds, the majority, hybrids that they've used, whether they have irrigation or not, and they just hit "run". It runs. It gives them the results. It's a chart for the last 10 years. Now they can play with it and say, "Okay, what if I had done something differently?" And now, if they want to compare it to a different field, they could look for the second field. Just add it to that portfolio, compare the two fields within the software platform, and all these thing, they could do within five minutes. It's all in that.

Brian:  So the shape shape file isn't just a map of coordinates of like what's planted where? It has additional data, and so when that's uploaded, you're getting additional data points that then go into the software and is interpolated. Is that correct?

Dinu:  No. We don't get any data except where the field is. So, all we need from the farmer is, where is his field, and by the coordinates, I simply meant if you were to draw a geometry around a map, that geometry is identified by the coordinates of each point that connect the lines, and that's all we need.

Brian:  Wow. That's pretty fascinating. So, I guess I could draw a shape map over the land that I want to purchase or rent, correct?

Dinu:  Exactly.

Brian:  And then you can do a comparison. Like if I run the same crops in the same... I know a lot about cotton. I don't know anything about eggplant, so I'm going to like run the simulation. Can I plant my cotton and bring my operations over there without making a... I assume they don't want to make dramatic changes to tools and stuff that they know. They want to stay in their own domain unless there's a big opportunity, or maybe they do.

Brian:  Maybe they say, "Wow, I can't plant cotton there, but you know it would be ripe for whatever, corn." Is that kind of the types of feedback you might be able to get?

Dinu:  It's possible. At this point in time, we only do corn, in the corn belt, so it's really about whether I want to irrigate, whether I want to plant more density of seeds. Which part of the field should I plant more? Which part of the field should I plant less? Should I have irrigation, or is it not worth the money? Things that kind of decision. Which hybrid should I plant? We capture that by what we call relative maturity. Those kind of information is what we allow the interface to change.

Dinu:  Well, let me frame this in a slightly broader way. Okay, there are a couple of things that the farmer is looking for. One is, they're looking for expanding to new pieces of land, so they're looking for which parts of land can be rented. But the farmers are also and that's the way the first one is a strategy to grow by growing your operations to larger areas of farmland.

Dinu:  The second strategy for growth is for new crops or specialty crops or organic farming. That's operational efficiency method of growing, and we have been talking about farmers and focused on the farming outcomes, but I initially started off this conversation by saying, CiBO is set up to actually serve the last-mile providers for the farmers, so I still want to make the connection that if the farmer likes what you're doing in terms of your technology, I think the last-mile operators' interests will be served in terms of their use cases.

Brian:  Have you had any, and maybe you haven't gotten this far with it yet, but have you been doing these kind of ride-along-style studies with the different personas? So not just the farmer, but say, the insurance person as well. Have you been doing similar research on the product?

Dinu:  We've done. We've done some research, not as much as we have with the farmers. But some with like I said, with agronomists, we have our own agronomists, and we work with other agronomists as well. We have done with ag import companies, in terms of seed breeders or product marketers for new seeds, so they're a community of different potential users that we work with.

Brian:  What are you finding difficult from a product management perspective? You talked about focus is really important. What are some of the challenges that you have in figuring out what should go into this tool, and is one of those challenges, like is the same interface for everyone the right product, or is it really, we need separate tools, separate products for each of these different personas? Even though they're all in the same ecosystem, they don't all need the same product. Does that I don't want to feed challenges to you, but can you talk a little bit about what's difficult in figuring out how to design a good product for these different groups?

Dinu:  Yeah, I think it's a great question. So, there are different levels of subject matter expertise, if you will, of the end users, all right, in the different verticals you might want to go after, and one of them, let's take an example of seed breeders. High expertise in data analytics. They want to see a lot of data and make sense out of it, and the best way to serve that particular community, because they come with their own philosophies of how to make these decisions, and the kind of decisions they're trying to make is, "Where do I test a particular breed of seed? Where do I best sell a particular breed of seed where it's going to be most effective?"

Dinu:  And the way we approach that market is by providing APIs to them and not an interface at all, an API that accesses the full power of our data engine and our simulation engine, so they can access the data they need and do analytics on top of it to help them make the right decisions. At the same time, once you get to the last mile, to the farmer group of companies like insurance providers or bankers or retailers or agronomists, each persona has a different use case, and the UI or the UX is designed, on purpose fit, to allow them to get the value differential value for the current workflow, which is going to be different from the other vertical.

Dinu:  That retailer's use case is going to be different from an insurance provider's use case. They're basically looking at different aspects of farmland or farming operations, and the way you do that is by designing the software architecture in a way that a platform is able to communicate with the user interface layer through APIs, and now you can make your user experience a very modular user experience with the kind of dashboard that collectively gives the information that is most relevant to who you're serving, and you can custom fit the user experience to serve that particular purpose.

Dinu:  So, yeah, we just change it. Each user experience is going to look different, depending on which target segment you are designing it for.

Brian:  And will those be separate products, or they'll still be one product, just with different modalities of use?

Dinu:  We don't know yet. Right now, we have what we call two products, one we call the CiBO Insights product, which is the application with one user interface, and the other is the CiBO API, which is for the subject matter experts in the ag import space.

Brian:  Right. So, in the original question, when we asked about what's difficult, are you saying is it difficult to figure out whether or not they should be separate products? Is that what you were saying?

Dinu:  No. It's easy for us to figure out if they should be separate products. What is difficult is the specific pain points you want to solve and how you expose that with a user experience that's intuitive.

Brian:  Sure, sure. Do you run the risk of too much data or too much of the raw information there clouding the insights and the value that you want to convey? Are you worried about that, or more like the number of different things you can do with the tool type of a problem? What are those challenges like?

Dinu:  I think the primary worry is to touch on the real pains that people have, right? The hardest thing for when you explore market needs is to analyze the market objectively and not frame the conversation to lead to the conclusion that you want to get to and truly understand what the current workflows are and what the pains are associated with the current workflows that you can provide a solution to and what the differential or the delta value of that solution is going to be. That is a big challenge.

Dinu:  The second one, of course, is what you said is, there is a lot of data, and you want to purpose fit your user experience to provide the solution you're targeting so the insights are clear and actionable and not confuse it with noise.

Brian:  Have you had any feedback that that's been the case so far, and you're like, "Oh, we changed X, Y, and Z report or whatever for this farmer because we found out that they didn't really need that information," or, "It wasn't the right stuff at the right time"? Have you, are you going through iterations and seeing these types of little learning outcomes?

Dinu:  Yeah. In fact, our approach was different. We exposed the minimal data that we thought was necessary, and we are actually going in the other direction, is to start exposing more data as we have more conversations, and I think we'll still hit the problem of, "Okay, now there is too much, and we need to retract," but at some point, we have a tipping point. At this point in time, our experiment has been very sparse data but not enough insights.

Dinu:  Let's provide for example, "Okay, you can run the simulation. You can tell me the last 10 years, how my field did, but I don't know what hybrid you used. Can you show it to me in a table? And when did I fertilize, and how much?"

Dinu:  So, suddenly now, we have a table that actually shows that information. Very soon, we'll have a situation where, "Help me compare the different fields against each other with the parameters that I want to compare them against and not the parameters you've already used." And so maybe we'll develop a custom layer of, how do you want to customize that comparison parameters and stuff like that, but at this point in time, we are going in the other direction.

Brian:  Yeah. This is actually something, I love that you said this. You know at some of the talks I give at conferences, one of the points that needs to be made is that, and I think even Tough Deed likes to talk about this, but to add clarity, you need to add information sometimes, and the issue isn't always the quantity of the information; it's how it's designed, and I've seen this repeatedly where there are times when, if you properly add information and design it well, you actually bring a lot more insight there.

Brian:  You could literally add 100 times more information to a screen. Perhaps you give a time series information in the form of some evidence to help back up the conclusion drawn by the software, and you actually provide more insight there, so this kind of like, "Remove stuff all the time." Well, not all the time. Sometimes you need to provide that evidence there, and I often talk about this pyramid, right?

Brian:  You have your conclusion at the top, which maybe is something as simple as a sentence. There's not even any data there. It's just a conclusion that the software interpolated, right? And then the next tier down might be that evidence layer. "How did it go? What did we look at," right? "What were the inputs? What are some of those factors? Why did we come up with the conclusion? Why this corn and not that corn?" And then the level three stuff is more of the data, right? All the insights and the goodies there, and what I've seen over time sometimes that happens is that over time, the user starts to trust. If they've seen it and they can believe it and they can relate to it, they start not to need the level three stuff as much because you've given them level two, and mostly, it's about a level one with that level two as kind of a check to help them believe in what's there, and that's all about the design, right?

Dinu:  Yeah.

Brian:  It's all about how you design that experience to such that they can believe these insights and get the value out of your analytics, so I love how you said you're actually on the, started small, and now you're gradually adding to provide that level of believability, and that's required.

Dinu:  Yeah. I can't agree with you more on that statement, Brian. In fact, we got to this point in the journey because we decided to separate the data-rich experience that a particular segment needs from the insight that the other segment is looking for.

Dinu:  We identified the ag experts that are likely to use our software will need access to a lot more data, and the way we'll enable is that is by APIs, and we'll reserve the app experience for specific workflows that are purpose fit for specific answers.

Brian:  Right. Right. No, I can see that.

Dinu:  Yeah.

Brian:  It's kind of funny how you have to add sometimes to make things easier, because a lot of times, we all assume that we need to be subtractive in our methods, but we just have to be careful with how we design it if we're using additive measures.

Brian:  So, it's interesting, and it's something, I think, as people that work in data product space, that we need to remember that sometimes the solution is going to be to add information, and if you're feeling like you have a clutter problem, if your customers are complaining about too much information, or that's a symptom usually that the design is wrong. It's not necessarily that that data has no value. It may be the wrong data. At the time they need it, say harvest time or whatever, you don't need information about what seeds to plant. It's not wrong information. It's just not needed at this particular time, so usually those are design problems lurking there when people talk about quantity. Whether it's too much or not enough, it's usually a design problem, I find, and then obviously, there is the challenge of actually building out, especially when they want something you don't have, you know data that you don't have. That's always a challenge, too.

Brian:  Are there any other particular challenges you guys are... ? This is really fascinating to kind of hear about these different use cases but especially because you've got these different stakeholders with different lenses on the whole end-to-end workflow, but any other particular challenges you guys are having?

Dinu:  I think we covered most of the parts, yes, Brian. The one thing I want to leave you with is how interesting ag space is, you know and that's one of the challenges we need to solve is not necessarily to make it boring but to keep the focus on the problems you want to solve.

Brian:  Sure. It does sound like a really interesting product. You know I'm really curious about it even though I'm not necessarily a farmer, but it sounds fascinating that that can be done. That's, I think, one of the fun things about working on data products, right, is there's so many different industries doing really creative things to solve problems for people, and I think it's fascinating, especially when you guys can do this without having... it sounds like you don't have a lot of data coming directly from the user, but they are able to get quite a bit of value out of the product because of how you've aggregated data from other sources and put it to good use and are providing experience around that that can help them produce food and crops and things that we need, so...

Dinu:  Yeah. Exactly.

Brian:  Awesome. Do you have any, I'm curious. This has been a great conversation. Do you have any closing thoughts for other people working on these data products and analytics tools that you might, some parting words?

Dinu:  I can't summarize that better than what we already had a conversation about, Brian. Sorry.

Brian:  Cool. No, not at all. No, you've shared some good stuff here, and where can people find out about... ? Obviously, I'll put a link to CiBO Technologies. It's cibotechnologies.com. Are you on Twitter or LinkedIn? Is there a place people can kind of follow what you're doing in this space?

Dinu:  Yeah, for a guy who's working on technology, I am less tech-friendly in that way. I only have a LinkedIn account.

Brian:  Oh, okay. Well, that's all right.

Dinu:  You have a fairly active LinkedIn account, and I see you post a lot there, and I enjoy some of the blogs that you write up.

Brian:  Oh, thank you so much. Yeah, I appreciate that. Well, I will definitely put a link to Dinu's LinkedIn page, and I appreciate you coming on the show. So this, again, Dinu Ajikutira from CiBO Technologies in product management, talking about ag tech. So, this has been great. Thanks again for coming on an experiencing data and sharing some of your insights.

Dinu:  Thank you for having me, Brian.

Brian:  Yeah. Take care.

This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

Subscribe for Podcast Updates

Join my DFA Insights mailing list to get weekly insights on creating human-centered data products, special offers on my training courses and seminars, and one-page briefs about each new episode of #ExperiencingData.