De Practicante en JP Morgan a Crear La Primera IA Que Desarrolla Alimentos: Vale $1.5 BILLONES!

It is not an evil industry. It tries to do the best with what it has. The problem is that what it has is not enough.

And we worked with LLMs and generative AI in 2015, before ChatGPT and OpenAI.

No idea is created alone, and I was with Nobel laureates.

And there are people who hate what we do, really.

I think for the startup world, I have to go quite far back. I think people might have seen a lot about NotCo, but they probably haven't seen the two previous ventures I did before NotCo. So my introduction to the startup world was even earlier, back when I was a finance bro, let's say. I was doing a banking internship at JPMorgan. So my background was always finance. I was born into a banker's home—my dad was a banker, so I heard banking language from the time I was very young. I would wake up to Bloomberg at 8:00 a.m. because my old man was watching the news. So that finance-related childhood was super interesting. I went on to actually study economics, and then I went into the banking system at JPMorgan in Hong Kong for an internship in private banking.

The interesting thing, and I think my first connection with entrepreneurship, was right there. One of the families we were helping with financial planning was an entrepreneur, and all his capital was generated through an enterprise. I got very involved in listening to his story, and it was incredible. It was one of those moments where your eyes light up and you say, 'I want this in my life.' The story is very entertaining: he was an Italian guy who had built the fifth largest yacht company in the world. The guy studied industrial design, and his father was a fisherman, so he focused a lot on making boats so that his father would at least have a more dignified trip out to sea every day. At the end of his degree, his graduation project was making a fishing boat for his father with a special design. He built it, and indeed his father went out fishing many times.

On one of those trips, an English millionaire who was passing by the coast of Naples told him, 'I love your boat. Who made it?' And he said, 'My son.' The guy gave him the number, the son called him, and the millionaire asked him to design a motorboat for him. That was how his boatyard started, and that was how he became who he was, having a spectacular net worth, obviously. But I think what caught my attention the most was how his eyes lit up when he told me the story of how he managed to do what he did, and that he was a happy guy with incredible energy. In banking, one lacks that spark when telling the story of, 'I buy cheap and sell high.' It's not very exciting. So seeing that, I said, 'I want this in my life.' That was literally the only corporate job I ever had in my life.

After that, I jumped into entrepreneurship, and I jumped into a path always linked to food. I always had a passion for nutrition, for physical exercise, for the body. I was always very interested in this industry. I always felt it was in its infancy, very primitive, very obsolete, zero science, zero technology. So I started there in Latin America, somewhat copying the Fitbit model, which at the time was the Whoop of today. I tried to copy that; it lasted a year and a half, and it went terribly. The truth is, I didn't know how to build a team, I didn't know how to delegate, I didn't know anything about the venture capital world, and I failed after a year and a half. It was crazy because it wasn't that I ran out of money—I just didn't know how to spend the money. You can imagine, totally the opposite problem.

Then came my garage story. I started a food company with a friend, making formulas at my friend's house, buying machinery in China and installing it at my friend's house in his sister's bedroom, who had just moved out of their parents' house. We developed a product and a brand that we got into the biggest supermarkets in Chile. We were the ones producing, labeling, delivering. We went to the supermarket, we were the shelf stockers, putting products on the shelves, monitoring sales every day. It was super interesting because, in the end, it gave me that perspective of the real garage startup world, where every peso matters, and of how to build a brand. But at the same time, it made me discover many things.

In that company, both of us had been university classmates and had studied economics, so we had nothing to do with the food industry. We hired an R&D firm to do the formulation for us. That world opened up to me for the first time, and I saw firsthand the technology and science used in the food industry to determine the products of the future, and I almost fell flat on my back. It was really three guys in white coats in an experimental kitchen doing trial and error, reading combinations of ingredients to generate certain flavors, colors, textures, and aromas. Clearly, with technology and science like that, we couldn't expect much from the outcome, and that was indeed the case. We paid for a year of development to this company—which, by the way, formulated for many other companies—and we ended up with a horrible product. In the end, we had to sell a horrible product to the market, very far from what we thought we were going to make.

Even so, with a really mediocre product, we were able to achieve spectacular placement in Walmart and Jumbo in Chile and have something going. At that moment, I knew right away that the transformational shift in this industry was not going to come from a new brand or a new concept; it was going to come 100% from a technology perspective. So after two years of being at this company and doing very well, I sold, and I went to the United States. There, I continued my journey of knowledge and scientific thinking, without ever losing this passion of, 'Man, it is possible. If someone hasn't done it before, it's not because it can't be done—it's because they haven't known how.'

Look, I think the most important thing is for people to understand that entrepreneurship is extremely hard, and passion is never enough. Passion falls short; passion is short-term. Passion is like setting fire to a newspaper—yes, it has an incredible flame, but it lasts very little time. So I think there's this myth of, 'Do what you love and you'll never work a day in your life because your passion carries you.' That is a lie. So I think the first step is to debunk the idea that passion helps generate this utopia of never having to work for anyone. That's false. Board meetings happen, problems come up, you go to work and have to solve 25 billion problems to find a solution, and it takes a ton of time. It's collecting payments from clients, things that don't work, a discontented supplier, a product someone doesn't like. Man, it is extremely hard.

At the same time, because you are generally doing something new, there aren't many people who can give you good advice. So finding those people very carefully is fundamental—surrounding yourself with people who complement you. And that's where I connect it a bit with therapy: it is imperatively important to know yourself before understanding what you need to look for outside to complement you. Therapy helps a lot with that. I was born in Chile and moved to live in Argentina. Argentina is perhaps the most therapized society in the world, and that started giving me a lot of resources around self-knowledge to understand what's happening out there—understanding why I do the things I do, what things I hate doing, and what kinds of people and profiles can do them much better than me.

So I think what I always had as a leader was this self-awareness, which is ultimately a superpower to understand very clearly where I need to focus my mind and where I need to put other people's minds, and whose minds I need to bring in. So therapy is good. It won't save your life, it won't solve the problems you have in a startup—I'm not painting it as the solution for everything, not at all—but it does give you a much better chance of going through this process, living it happily, and being able to endure it, because enduring 10 years of startup life is extremely hard. It's very hard.

Exactly. Exactly. I think what sticks with me, and what really helps, is understanding your purpose, because purpose is very different from passion. Purpose is an underlying asset of yours that must be very clear and very present at all times. That is the cushion where, when you fall, you actually bounce back. Passion is not that. Purpose is fundamental to me, along with maintaining authenticity. I think what therapy allows you to do is just that: to be okay with who you are and work on some things you might want to become, but staying authentic.

Authenticity helps you. For me personally, when raising funds—going to Andreessen, going to Sequoia, seeing Doug Leone, dropping into meetings unexpectedly—knowing how to show up as myself and not like the other 10 guys who pitched before me, I think that has also been fundamental, and that is largely worked on in therapy.

That's a great question, and the short answer is no. It is not an evil industry. It is an industry trying to do the best with what it has. The problem is that what it has is not enough. It's not enough for the world we have, it's not enough for the world's population, it's not enough for the resources we have left, it's not enough for coexisting with other species on the planet—it's not enough. So we shouldn't blame the food industry for being evil. I also started out with that stigma, thinking there are evil people making evil decisions, and it's not like that. It's really someone you can't blame because the tools they have to decide the future are also very scarce.

That's where the creation of NotCo comes from. I realized this: I realized that the tools and the science behind food make them make very bad decisions, both environmental and for human health. There is massive ignorance. But we also have to be fair: in terms of the information available to us as consumers in the past, it was zero. As a kid, it was impossible for me to relate the piece of steak on the table to climate change—I had no idea. I had no idea how related processed meats were to the potential onset of certain cancers or any of the four horsemen we will all face in life. The four horsemen are basically metabolic syndrome, heart disease, cancer, and cognitive decline—Alzheimer's or Parkinson's, etc.

So today it's already there, and as we grew up, it started coming to light. What you mentioned is great: Cowspiracy. Cowspiracy taught us things we had never seen—Cowspiracy, What the Health, The Game Changers. Information was coming out that would obviously change the way consumer packaged goods decisions were made. What there definitely needed to be was a tool that would allow people to eat better without even realizing it, because we are human, we like to eat tasty food, we don't want to change things, and we say we eat well when we really don't eat that well. So there is also a ton of knowledge about human beings that is fundamental to understanding what needs to come next as a product, what we have to launch. That was the concept: technology, technology, technology to make something completely new for what the future holds for us.

I would tell you there are two very distinct dimensions: one is the consumer's perception of what is good and what is bad, and the other is what is truly good and bad according to science. These are two things that often don't align. So I'd say we have an industry where all the information it generates is very confusing. One day eggs are good, one day eggs are bad. One day milk is good, one day milk is bad. One day cheese is good, one day cheese is bad. One day public enemy number one is calories, but then it's saturated fats, then cholesterol, then GMOs, then gluten. It's an industry that seems unable to agree on what is truly good and bad, and that confusion gets amplified a lot. Myths start forming around food that aren't necessarily true; they are very far from it.

Let me connect this a bit to the downside of social media, where we often idolize pseudo-scientists who aren't even scientists, and they say, 'No, seed oils cause inflammation, blah blah blah.' You look at the literature, and it's completely the opposite: they are anti-inflammatory. Then you go to processed foods—sure, but engineering is what allows you to create nutrition that you would never find in a fruit or a vegetable alone. So the nutritional value of a processed product is extremely good. Now, they put everything in the same bucket: a processed product is indeed a Twinkie or a cookie that is very bad for you, but that's not the same as a protein bar that provides all essential amino acids, which may have 10 or 15 ingredients, but was engineered to be a much better product than what you'd find in nature.

So there is this dichotomy between what the consumer thinks, what influencers proclaim, and what science says. At NotCo, what we are trying to do is bring those two worlds together. And there are people who hate what we do—they really detest it, and that's fine. It's crazy with those people, but when they try the product, they say, 'I can't believe it.' There are critics who, when you give them a burger, a NotBurger inside a bun, they say, 'How great is this burger!' And when you tell them it's a NotBurger, they say, 'I can't believe it! If you had told me, I would have never eaten it.' There's also a mythologizing of products and things that doesn't match reality. So there is a huge gap between what we believe food is and what science says is or isn't good for us. Bridging that gap is important, and it's hard.

You mentioned something—I loved that you brought it up: Monsanto. If it weren't for Monsanto, we would be starving; the entire world would have nutritional deficits. It's controversial, right? Because I also think Monsanto's business model—appropriating IP for certain seeds that enabled better yields, convincing farmers, and so on—was very poorly thought out. Monsanto's science is insane, which is why Bayer bought it. So we could discuss this for a long time, but what I'm saying is that people might criticize Monsanto, but they were eating Monsanto. It's the same as Coca-Cola: people criticize ultra-processed foods, but they drink Coca-Cola the next morning.

You see trends toward everything being more organic, more functional, better, but that's not the whole story—all Coke has done is increase its market share. So there is a perception, and there is a reality. This is the same thing, and it transfers very humanly to what is happening in politics, for example. There are things that are politically incorrect to say, but people do them anyway. I think there has never been a greater disconnect between polls and actual voting because there is also a public penalty for what one thinks. Nutrition is somewhat like that.

In short, they don't know how to do it. For them, you have to think about something that I didn't understand at first: think about when was the last time a big company launched a product that created a multi-billion-dollar brand. I have the answer: it was 50 years ago. Large companies struggle to innovate. What they prefer to do is fix tiny things in extremely large existing products, because improving gross margin by 1% for Oreo, Cadbury, or Nutella is much more important than launching a new product. That one point of gross margin generates $500 million or $3 billion in absolute profit. That is a major issue, and it's because multi-billion-dollar food giants prefer to promote big existing items rather than continue creating small new things for the future.

Now we are at a point where they actually need a bit of this side: how to transform their company to do more things like what NotCo did. Artificial intelligence is difficult to implement, difficult to work with, and difficult to understand. NotCo was born in 2015, and we worked with LLMs and generative AI back in 2015, before ChatGPT and OpenAI. The great complexity, the great challenge, was first in understanding food. Food is very difficult to comprehend because it is multidimensional. We have to view a formula as a multidimensional problem. If I remove, say, 5 grams of sugar, I have to complement it with something, but that something is not directly proportional. Swapping one chocolate for another chocolate, or one cocoa for another cocoa, is not directly proportional—it changes absolutely everything.

So what was the first thing we had to do? Understand why humans eat the things we eat, why we like the things we like, why we prefer one brand over another. We had to dive into the relationship between the molecular components of food and human perception of flavor, texture, aroma, and color—and understand that there are also many differences between human beings. What I perceive is not the same as what you perceive. If I was born in Mexico, it's very different from Ecuador, very different from Chile, very different from Argentina, and immensely different from Asia and Europe. They are completely different perceptions, and those perceptions are also linked to feelings. It wasn't just the molecular data of the food; it was human perception of many things.

So we had to start connecting the dots. The first thing NotCo did was develop this intelligence around what data we needed to look at to understand what a product is for me and for you. We went to spectral data, MS-Spec, GC-MS, infrared, and a bunch of things that, once again, had never been used in the food industry. When I left my last startup to create NotCo, I went to Berkeley to study as much science as I could, and I got very close to the biochemistry department. I saw that 99% of that department was working in pharma, and pharma was doing everything the food industry needed. The issue was that R&D wasn't being paid for in food because pharma margins are much larger than food margins.

So it was about taking the nuances of what was happening in pharma and bringing them to the food industry. That was how NotCo started with this more scientific, technological approach. Once we understood food and why we liked the foods we consumed, the question was: how could we make exactly the same things we loved eating, but in a better way? And when we asked science what that better way was, it was to make everything from plants. What other challenge was there? There are more than 300,000 plant species in the world, and we don't know what they can do. We really have a repository of less than 1% of total plants to understand what they can do in terms of flavor, color, texture, and aroma.

So for us, it was about exploring this completely unknown plant world. We didn't know if the combination of cabbage and pineapple would generate the flavor of milk, or if the combination of tomatoes and peach would generate the flavor of chicken. Maybe for a human being it would have been impossible to think that would actually happen, but not for us. An artificial intelligence that is unbiased and understands that the chemical reaction between these ingredients—in the case of cabbage and pineapple—generates lactone, making you feel like you are consuming a dairy product, that is the power of artificial intelligence. It's understanding how to replicate the same thing with other ingredients that might be better for the world and better for human beings. And that's how we started.

We knew, however, that if we wanted to build the technology that could become the 'Intel Inside' of the entire industry, we had to pave our own way. We had to launch our own brand, we had to make a mainstream product, and that was the challenge. Well, six years later, here we are. That was a bit of how we thought technology was going to change absolutely everything, and how we replicated what worked in pharma and brought it into food.

Exactly.

Clearly not. How did this idea come about? I think this is new. I did two internships: one at LarrainVial, which is the largest asset management firm in Chile, and then at JPMorgan. At LarrainVial, I worked in the business intelligence department around 2010 or 2011, and those were the first notions of artificial intelligence I had in my life. Instead of using artificial intelligence for food and much more complex things, I used it much more to build the best mutual funds and things of that nature. So K-means clusters and a lot of AI language—I started picking it up there, and I said, 'This food industry looks quite a lot like the financial industry in many aspects.'

I went through the 2008 subprime crisis. I saw what was happening: there were large companies selling hyper-complex products to consumers who were totally disconnected from what they were buying, and where the regulators became the regulated. We had those same four things in the food industry—the exact same ones. So I thought, with this approach I learned in the financial industry, there might be something in the food industry that I can change. That didn't make me a scientist or anything like that. I met with Nobel laureates at Berkeley to bounce this idea around, so it's not that I was the scientist, not at all. But I was in close contact with many scientists to understand the path, what had been done, and what hadn't.

I didn't reach this alone either, and that's important: no idea is created alone. That's also a good message, and ideas are worth zero if they aren't executed well. So I looked for scientific co-founders. Karim is a PhD and postdoc from Harvard in computer science, working in astrophysics. The interesting thing about the story with Karim is that many of my classmates who studied engineering had Karim as a professor, and everyone spoke about this professor who knew artificial intelligence like no one else in the world. It was legendary because his classes were packed, with people looking through the windows to watch Karim's lectures. I said, 'I have to meet this legendary figure, and obviously I have to bring him this idea to see how AI can help me rebuild the food industry.'

When I spoke with him, I obviously thought there was zero chance of capturing his curiosity, and it was crazy because what he did was extract data from telescopes and develop AI algorithms for astronomers to understand the composition of a star, atmospheric density, and ultra-complex things. But what he told me later, once we co-founded NotCo, was, 'What interested me most was that everything I do, I can't test. I can't go to a planet five light-years away and prove the atmospheric density, the elements, etc. In food, I can do it, and I can do it immediately. It's something very tangible that I can change on the spot.' So that caught his attention tremendously. He bought into the whole vision, and we went for it.

On the other hand, the other co-founder is Pablo. Pablo is a biochemist, PhD in biotechnology, and an expert in plant genomics. Pablo was the mad scientist who had spent three years going to Antarctica, analyzing how plants could survive in these extreme climates. So the scientific pillar came from the scientific curiosity I always had, but also from the scientific complement I needed to have in place to achieve what we achieved.

Technology, to me, was fundamental to changing the food industry. What I saw was that it wasn't an evil industry; it was a very ignorant industry—ignorant of the technical capabilities that existed to revolutionize the industry they were leading. Now, I knew it would be difficult to change ways and change technologies internally within companies—almost impossible. There was also a lot of human friction: humans don't want to be replaced, and many people see artificial intelligence as a replacement. But if they see that commercially, by using this tech platform, you are doing things they can't do, that sparks curiosity, especially if you are taking market share from them and they are suffering in the market.

So NotCo was always conceived as a technology company where the way to monetize that technology would vary according to the path we built. And this is super curious: the first invoice NotCo ever issued was to another company. We wanted to become an R&D company for the CPG industry, and it didn't work out for us because what we envisioned was very different from what other companies envisioned. The things we wanted to do at the point of sale were completely different from the vision and values of those other companies. That was when we said, 'Wait, if we want to build the Intel Inside, we have to create the iPhone.' It sounds crazy, right? It was like, 'How can we pave the future path ourselves, even if it costs us 10 times more to build a brand, build a CPG business, and fight for shelf space in multiple countries?' But let's go. And that was the start.

The first time we thought about launching our products, we took two years to develop the technology before launching the first product, and that first product was NotMayo in Chile. There are things that, let's say, are interesting. I don't know if you've been to Chile, but Chile is the third largest per capita consumer of mayonnaise in the world. You go to a supermarket and there's a whole aisle of mayonnaise. It's crazy. There isn't any other sauce, but it's full of mayonnaise. So we launched NotMayo, and people hooked onto the product right away because it wasn't that they were waiting for a new mayo; they were waiting for a very new, very disruptive way of looking at the industry that spoke to what new consumers were thinking. We achieved a 7% market share of the mayonnaise category in 10 months of sales. And that was when Kaszek saw this data, saw our disruption, and they were like, "Wait, wait, tell us a bit more."

In December 2017, we received our first investment from Kaszek, which is the largest venture capital fund in Latin America, former founders of Mercado Libre, etc. And that's where our journey started: it was NotMayo, IceCream, Burger, Chicken, Milk in Chile, Argentina, Brazil, Colombia, Mexico, the United States, and Canada. We raised $420 million. We started working with incredible companies like Burger King, Starbucks, Papa John's, Shake Shack. And for the first time, for example, we became Starbucks' innovation partner. We introduced food, we introduced the milk. And notice that Starbucks started experiencing a tremendous change in consumption. Today, just to give you an idea, 78% of total beverage sales in Starbucks stores are cold, not hot. 78% is crazy, all year round. So it's not even a seasonal number I'm giving you.

And that complicated things for them because the products they offered, the machines they had internally, the profit of their products completely changed. And one of the things they couldn't do cold was foaming. When you buy a cappuccino at Starbucks, you buy 40% air—expensive air. Cold, they couldn't do it. And they came to us with this problem: "NotCo, with your artificial intelligence, can you develop a product for us that foams cold extremely well?" And we said, "Why not?" And that's how it went. In nine months, we had a product that was selling in more than 3,000 locations, which generated business incrementality, better profits, etc. And just as we did with Starbucks, we did it with many other fast food or consumer chains.

And it was like, well, we are already being this innovation partner, we are using our technology and artificial intelligence to develop and drive incrementality in the business of these kinds of companies—why are we limited to these companies? So we launched in the United States with NotMilk, NotBurger, NotChicken. We did very well, and I received a call from the global CEO of Kraft Heinz, Miguel Patricio. And Miguel Patricio tells me, "Mati, I want to do something meaningful with NotCo. We love the brand, we love the speed, we can't believe the quality of your products. Why don't we make the 'Not' version of our traditional products: Mac & Cheese, Kraft Singles, Heinz, Oscar Mayer, Philadelphia?" We said, "Well, but it's important to understand how." And he tells me, "Look, Mati, what I want most is for you to lead it. If we do a joint venture, I want you to lead this and for you to appoint the CEO, because if it lives inside Kraft, this will fail."

And it was like, damn, I had never heard so much self-awareness from someone in the corporate world. And it was like, well, let's explore it. A year later, we founded The Kraft Heinz Not Company. We have eight product categories operating here in the United States, in more than 10,000 points of sale. It's going extremely well. And for the first time, we saw the realization of this vision we had eight years ago of being the 'Intel Inside' of the CPG industry. First it was Kraft Heinz, and now there are a ton of new updates that we can discuss. So that was the path.

We never would have been a billion-dollar company if we had followed that path, never. There were two companies that offered to buy our technology early on, but we knew that the objectives for using that technology were going to be completely different from ours. We had to demonstrate a way of doing things that was different from how other companies did it. And not only that, what's crazy now is that they not only ask us for the technical solution to some brand problems that large companies have—like sugar reduction or cost reduction, etc.—they also ask us for help with marketing. It's crazy. So part of the service we provide is almost an innovation and marketing service, because the industry supporting CPG is also very outdated. Today, for example, I think NotCo has done very good marketing; it's very catchy and very disruptive, and we talk about things others don't, so we connect with consumers in ways other companies don't.

Through this, we realized that if we really wanted to help these big companies, we also had to understand their internal pain points. The first is what you mentioned before: why couldn't they do what we did? Why not, if they are multibillion-dollar companies? The first reason is that they don't have the priorities, they don't know how, but they also have a massive amount of data—an infinite amount, I tell you—decentralized, very difficult to read. So the value they are generating with the data they have is close to zero. So before we get into the deep technical side, the first step is: let NotCo help you centralize data and generate a lot of value with the data you already have for your internal processes.

The second thing is: let me fix your innovation process. In general, large companies hire marketing agencies to produce innovation concepts. They make incredible designs, test them with consumers, give you qualitative and quantitative data, but then they tell you, "Here, this is the validated product, you make it." It's another completely broken process, because what marketing produces, it hands over to R&D. R&D has to guess what kind of product goes inside that validated concept. And the marketing agency has zero accountability for the success or failure of the product on the shelf. So their job is super easy.

We said, "Well, let's create a technology module that helps them generate the concept, understand that there is a white space, understand the opportunity, and we develop the product for them, being partners and associates in the launch of that product." So if it's successful, we both do well; if it's a failure, we both do poorly. So we also changed the operating dynamics of this industry with artificial intelligence. If we started with this idea of selling a concept and service to CPG companies and it didn't work, we went out and did eight years of commercial groundwork showing them how things were done, in order to return to the same place we were six or seven years ago, but obviously understanding in depth the pain points they had. Because what kept them up at night was the same thing that kept us up at night as a CPG company. So that was a virtue.