How Crowdbotics Helps InnerPlant Revolutionize Farming with Synthetic Biology
Santiago Freyria is the head of product at InnerPlant, a startup that codes signaling capabilities into the DNA of crops to detect threats within 24 hours of emergence.
InnerPlant + Crowdbotics
Santiago Freyria knows what it takes to build a great product. As the first product manager and second overall hire at Clip, he played a key role in growing the company to over 200 employees and a nearly $400 million valuation. In his subsequent tenures as a consultant ( BCG), investment analyst ( Bee Partners), co-founder ( Callisto, StEP), MBA student (UC Berkeley ‘20), and now head of product at InnerPlant, his passion for product execution has only grown.
So when he approached Crowdbotics to build InnerPlant’s MVP, he knew what he was doing. “I received very positive feedback from a few startups that had worked with Crowdbotics, and once I saw firsthand how Crowdbotics operates, I understood why.” For Santiago, speed was everything: “Crowdbotics pretty much saved my life by building our product super rapidly and enabling me to focus on getting traction fast.”
InnerPlant codes crops’ DNA to enable them to emit fluorescent proteins when they experience stress. These visual signals can be read remotely from as far away as satellites, giving farmers an early detection system for specific field threats such as insects, fungi, diseases, changes in weather, water deficiency, nutrient deficiency, and excessive watering. We recently sat down with Santiago to discuss his experience with Crowdbotics and his larger plans for InnerPlant following initial product development.
Crowdbotics: So, just to give a quick overview, what is your background, both professionally and from a technical perspective?
Santiago: I studied economics in Mexico, and I was very idealistic, you know, trying to fight social inequality. But I realized that I could do that way better from companies-specifically from startups-that will actually make changes, than from the government, where I only found corruption, politics, bureaucracy, and so on.
So I joined Clip as the second hire. Clip is a startup enabling people to process and pay with cards, and it is now the largest fintech startup in Mexico, having raised $100 million from SoftBank two years ago. After two and a half years, I left Clip and joined Boston Consulting Group (BCG) just to explore industries. I was also planning to get an MBA, and consulting is a pretty good path to do so.
For two and a half years at BCG, I worked across a couple of industries. Then, I got accepted into the Haas School of Business at Berkeley. I wanted to come to the Bay Area, so the day I got accepted into Haas, I quit BCG and joined a venture capital firm in South Africa-the second largest firm in the country-for three months as an intern. And there I ended up actually building a product on how to help entrepreneurs execute better, so I’ve pretty much been around product my whole life.
When I joined the MBA program, I became co-president of Berkeley Entrepreneurs Association, and I also launched StEP (Student Entrepreneurship Program), which is this startup incubator. Over a year, it became the largest incubator program on campus.
And while I was doing that, I also started another company with a couple of friends. It’s called Callisto. We do rum infused with California botanicals, and, again, I was very involved in the product. I was pretty much cooking rum in my kitchen every single night, which was pretty fun. At some point, we already had 2000 bottles done through contract manufacturing, designed with the flavor profiles that we had been exploring in the kitchen, and after getting enough feedback from customers. So, once the prototype was not there anymore, it didn’t make sense for me to stay there. So we raised over $600K, we grew the team, and I stepped down.
I was also working for Bee Partners. Bee Partners is an investor both in Crowdbotics and in InnerPlant, and what I was trying to figure out is: “With so many cool startups out there, how can I put on my investor hat and decide where I want to spend my time and resources?”
I saw over 2,000 companies, and one of the companies I was actually targeting was Crowdbotics, so I helped you with a couple of projects. I really liked Crowdbotics for a number of reasons, so I was thinking about exploring the opportunity of joining you guys full time.
But at the same point, I was very interested in synthetic biology. Then I heard about InnerPlant, and the vision of revolutionizing farming is fascinating. There are so many reasons I joined InnerPlant — we bring value back to farmers to ensure we can feed future generations. I joined as the head of product, and now I’m director of product. I’ve been working with InnerPlant for over a year already, and full-time since June.
Crowdbotics: Wow, that’s quite a journey. Would you classify yourself as a serial entrepreneur? It sounds like you weren’t necessarily founding all these ventures, but that you’re often the first or second hire, an early investor, or early advisor in a number of startups.
Santiago: Yes. I think I don’t necessarily come up with the big initial ideas. Even the companies I have founded have been based on somebody else’s idea, including both StEP and Callisto. What I really enjoy is the execution part of it. I like joining very small teams, where I have a lot of bandwidth and freedom to operate, and just jumping into executing.
Crowdbotics: Okay, that’s great. It’s not always the case that our clients have deep knowledge of product execution. Many of our clients are idea-oriented, with more of a traditionally entrepreneurial inclination. So, it’s interesting that you’re more experienced with the nuts and bolts of iterating on a product, growing it, identifying its strengths and weaknesses, and being involved in those high-impact early decisions. That’s historically a less common client profile for us, but it’s one that we’re seeing a lot more of.
So, by your count, how many software products have you built in this capacity, as a product manager or head of product?
Santiago: Software-wise, I oversaw the fraud prevention product at Clip, and I have been involved with the websites of Callisto and StEP using Squarespace in both, plus a bit of InnerPlant’s website, and, of course, the software product at InnerPlant.
Crowdbotics: So, you found us through Bee Partners, but what made you decide to choose us for building InnerPlant?
Santiago: It was mostly that I saw the way you operate, and I received very positive feedback from the StEP startups that worked with Crowdbotics, and Crowdbotics was also highly recommended by Bee Partners.
“I needed specific answers right before an investor meeting, so I reached out to my Crowdbotics PM. She jumped into a call that very second.”
Crowdbotics: Wonderful. It’s always great to hear that our clients recommend us.
Let’s get into InnerPlant a little bit more. If you could, please tell me about both the company and the product itself, and what exactly Crowdbotics has built for you.
Santiago: So, what we do at InnerPlant is we give plants a voice. Over millions of years of evolution, plants have developed different ways to interact with each other and protect themselves from the environment. Whenever stress affects a plant, plants release chemical signals to other plants, and they create proteins and defense mechanisms to defend themselves from that specific stressor.
It can be insects, fungi, diseases, changes in weather, water deficiency, nutrient deficiency, excessive watering. Whenever a plant is not in optimal conditions, it reacts in a given way, and it also tells other plants that there’s something coming so that they react in the same way.
So, what we do is we code plants’ DNA. We insert a fluorescent protein within the promoters that react to stress, so that whenever they have these natural reactions, they also emit a fluorescent protein. We have different colors for different stress types.
What happens right now is farmers have all these hardware solutions, and drones, and sensors, and satellite imagery, trying to understand fields’ health. And whenever they detect a change in biomass, or the field is already yellow, it’s already too late because your yield is already going down. At this point, you cannot reverse it. You can just control it.
What we do is that, as soon as the plants experience stress, they emit a signal to us that we can read remotely from as far as satellites. And we can tell what the specific stress factor is, so we provide farmers with recommendations on how to manage their fields. Whenever stress comes, we know it 12 hours after it’s there, while with external technologies, it’s probably like four weeks.
Crowdbotics: That’s fascinating. I’m curious to know whether, given that different plants have different physical structures, do they all emit similar stress signals? Or is it unique from crop type to crop type?
Santiago: If you think about it from an evolutionary perspective, plants share more than 99% DNA, so most of these things are the same. There are promoters that will be specific to different crops, but, for example, most plants will react to water stress the same way. If they have too little water, then they need to become more efficient in their water use.
Or, let’s say you have insects coming. In that case, they need to emit chemicals. If you think about a drought coming, then they need to produce a lot of fruit so that there are seeds, because the current generation might not stay for long, right?
“Crowdbotics took some basic Adobe XD wireframes and rapidly turned them into a product demo that we are now using to raise investment capital.”
So, if you think about it, these reactions are very widespread across crops. The component that we’re adding is a fluorescent protein that’s not native to plants. It’s a protein that plants don’t naturally produce, and we’re inserting that protein into plants’ DNA so that they can actually produce it and emit it.
Crowdbotics: Got it. So, you’re using devices and visual tracking to read the signals generated by this protein. What is the actual user interface for someone who is picking up these signals from a given field of crops?
Santiago: In terms of hardware, we can detect from handheld devices, drones, airplanes, satellites, and even from an iPhone. The in-app user interface is more of a software product that provides recommendations for farmers, and that part is enabled by synthetic biology and physics of light detection.
Crowdbotics: So there’s the basic detection tool that just uses a device’s camera, but then there’s also an in-app ecosystem of recommendations and communication.
Santiago: So far, what we have built has a heatmap, and a very visual way of understanding field threats. We also have a collaboration tool, which becomes kind of a farm management application by itself. And then we have all the actions that we’re recommending, we have an analytics tab where you can see a dashboard, then field health across time, and really just a one-stop place to see everything happening on the field.
Then we also have an ROI tab that actually talks about the financial impact of the actions you’ve taken. There are overviews of multiple fields, which is mostly a tool for agronomists that are managing multiple fields at the same time, and integrations with other applications such as weather forecast services. And that’s pretty much it as of right now.
Crowdbotics: And what platforms has it been built for so far?
Santiago: Right now, we’ve just built the front end, because we’re building it in a very agile way. So, I built some basic mockups in PowerPoint, and I got a lot of feedback from farmers, investors, and so on. Then I built wireframes on Adobe XD, collected even more feedback, and then I built this front end that shows all the features and functionalities. And I’m still collecting feedback.
Right now it’s web-based, because that was the most seamless approach. But eventually, we will push this into Android and iOS once we have both the front end and back end fully built.
Crowdbotics: So it’ll be a web app, then launch to mobile devices? Or is the full hardware strategy not quite mapped yet?
Santiago: No, I think we’ll push things into mobile devices very rapidly because farmers are not on their laptops that often. We just started with web because we needed to build something super fast that we could share with investors and use it to collect feedback from farmers in the most ideal way.
Crowdbotics: When it comes to conducting field tests, are you currently just looking in the Berkeley area for local farmers?
Santiago: We’re actually working on tomatoes this year, and most tomato farmers are in Central California.
Crowdbotics: Is there anything about tomatoes that makes them a good candidate?
Santiago: It’s a very easy crop to transform. It’s in California, which means we don’t need to travel to do the trials. There’s a lot of research into it, so it was our best shot for something to work, and it actually did work. And then it’s a high-value crop, which means demand is way higher, and it’s a crop with a lot of problems due to pests, diseases, and water stress.
Crowdbotics: Is there an optimal size of farm or farming operation for this product, or do you feel that InnerPlant can be scaled up or down as needed?
Santiago: I think the value increases with the size of the farm. If you have three tomatoes in your garden, then then you can literally take care of each plant. If you have 10,000 acres, then it’s very hard to manage those fields, right? You have 10 agronomists, and you’re walking all over the place, so I think the larger the farm, the more value we deliver.
Crowdbotics: What materials did you bring to us at the start of the build?
Santiago: I created some very detailed Adobe XD wireframes that mapped all the functionality and all of the components. I also brought a lot of design guides to capture the look and feel of the brand. And I brought a lot of materials designed by designers, like the heat maps, and I brought the content as well.
Crowdbotics: That approach is almost always going to result in a more successful build with us, where you’re going from design to development, you’ve already mapped user flows, and you’re coming to the table with pre-existing content.
Getting into the build process a little bit, are there any big wins or key features that you can attribute specifically to Crowdbotics? Something that we brought to the table as a unique contribution?
Santiago: There were three things, actually. The first one was that they didn’t just take what I sent them and start building. They actually challenged a lot of materials I provided and helped me think about how to make all of the flows more user-friendly. A couple of the team’s recommendations were very valuable.
The second one was speed, because they were always sharing new features, and I could react very rapidly. And the whole project was completed within a few weeks.
The third one was design, because they initially used the same designs that I had in the wireframes just to ensure that everything was working. That was a very agile approach. And then they invested into adding higher-fidelity design, and they had designers help me with a couple of screens. I pretty much just had some Tableau charts, and the Crowdbotics team helped me make them way more sleek.
Crowdbotics: More generally, what was your day-to-day experience like working with the Crowdbotics team? How did you feel as a client?
Santiago: I think the communication was pretty good. I remember a time I had an investor meeting and needed to have specific answers that I didn’t have on hand, and I just reached out to my product manager, and she jumped into a call, like, that same second. So the reaction speed is super good.
Crowdbotics: Did you feel like you had visibility into the build at all times? Were you more hands-on, checking in constantly, or did you prefer to just get updates from us on a regular basis?
Santiago: I was more passive, because we’re an eight-person team, where the CEO is fundraising, and I’m the only non-technical person on the team. So as I was doing product, I was also running field trials, doing marketing, and doing business development. So, I was pretty hands-off. I just trusted Crowdbotics to do what they had to do.
Crowdbotics: Cool. Sometimes we have clients who prefer to go in the opposite direction, where they’re very, very involved, and it’s day-to-day feedback. It’s great to know that you can also hand it off and just keep an eye on it and intervene only when necessary.
So it sounds like you’re taking this frontend and this current product demo into a round of fundraising here, but, more generally, what are the big upcoming mile markers for InnerPlant?
Santiago: Next is building the backend, designing the whole database infrastructure. We’ll be running field trials, so we need to understand how we’re capturing data, what types of data we’re actually integrating into the system, how the data tables connect to each other, and the relational components between them. We’ll be mapping the backend and getting closer to having complete functionality. As we do that, we’ll be collecting a lot of feedback from farmers.
I think in the beginning, we’re going to be leveraging Crowdbotics more, but eventually we want to bring the product development in-house. So we’ll need to plan that transition and determine just how much we want to take on ourselves.
“If you’re a startup that isn’t prepared to start hiring a full product or engineering team, I recommend hiring Crowdbotics to build a high-quality MVP as a foundation for your eventual product.”
The other part that’s currently missing is the whole data component. Once we start collecting a lot of relevant data, we’ll need machine learning algorithms and ways to interpret the data to provide valuable recommendations for farmers. And I think that one of Crowdbotics’ largest strengths is understanding how to build machine learning applications in a very fast way.
Crowdbotics: So you’re saying that, even if you bring a large portion of product development in-house, there may still be cases where you would hire Crowdbotics for one-off projects? Or do you see this approach as more of a way to use Crowdbotics to build a robust MVP and then hand that off completely to your in-house team?
Santiago: It’s more the latter. Specifically for startups, if you don’t want to start hiring a lot of developers right off the bat, or if you don’t have much of a technical background, it’s probably worth just hiring Crowdbotics, getting the MVP that you need, and creating a product and traction to get funding, even before hiring somebody.
Crowdbotics: I think we do see that use case pop up a bit more frequently with startup clients. We’re always happy to grow with a company and be a long-term partner, but every company grows in its own way. We aim to be flexible enough to assist with whatever a company’s needs are.
Santiago: Yes, I think that using Crowdbotics pretty much saved my life by building our product super rapidly, enabling me to focus on getting traction fast.
Originally published on the Crowdbotics Blog November 17, 2020.