Following their recent collaboration with Kraft-Heinz, Luke Wilkinson spoke to NotCo’s Rodrigo Contreras about the company’s game-changing AI ‘Giuseppe’, finding that perfect meaty texture, and how cabbage might be the key to a delicious plant-based milk.
The company came from an idea put together by the three founders. Their idea was to use food technology to develop plant-based food, starting from a very simple principle: all of the animals that we eat are herbivores - bio factories for turning plants into meat products which we eat. The question was whether the animals could be removed from the equation. This shared goal brought together Karim Pichara, an expert in machine learning and AI, Pablo Zamora, an expert in plant biotechnology, and Matthias Muchnick, a natural businessman and entrepreneur. They decided to work together on the project which later became NotCo.
At that time, the plant based food market basically didn't satisfy any of the needs of consumers in terms of the actual experience of the food they provided. One of the great drivers of NotCo was to make products that were both nutritious and delicious. That was where the opportunity lay: not only in appealing to veggies and vegans but meat eaters, too, by offering a better sensory experience so that they will happily opt for a plant-based option.
The consequence of a plant-based option like ours is both a better diet and greatly reduced environmental impact when compared with food derived from animals. That was really our starting point as a company but, in the end, we are not as much a food company as much as we are a food tech company; what we’re really interested in is creating technology that allows us to make even better products.
I am the manager of scientific research, part of the research and development department. R&D is a huge department and much of it is dedicated to the engineering of our products so that they meet the expectations of the consumer, whereas my team is a little more in the background, working on the understanding of the composition of the product. We also work on generating the data necessary to send to the artificial intellence team so that they can develop our AI, ‘Giuseppe’.
For me, biotechnology is the future of food. It's clear now that by 2050 the natural resources that we currently consider a right will become commodities - like water, for instance. We need the benefits of biotechnology in order to have a food production system that is sustainable and allows for the democratization of food; a food production system that gives people a better quality of life.
What we need right now is a more efficient use of our natural resources but, as things stand, much of the world’s animal protein is at a price level that is inaccessible to certain people, besides having a huge impact on the environment. People simply can't afford to eat healthily. This works against the FAO and the UN, who both say that health is a human right; if we don't address the cause of ill-health, i.e. poor diets, there won't be a way to give people that right they have to better health.
Biotechnology can be key to lowering food costs, and improving efficiency in the use of water and reducing CO2 emissions. All of this will also have an impact on current levels of environmental damage.
It changes from country to country but, here in Chile, we have Not Mayo, a replacement for mayonnaise, Not Milk, which comes in different varieties – low-fat, chocolate, et cetera - and the meats: Not Burgers, which are beef patty replacements in a few different styles: one for restaurants which has a slightly different size, one for supermarket retail, and one that we make exclusively for Burger King, which is called the Whopper. The Whopper is produced by different companies in different countries, for example in Europe they are made by The Vegetarian Butcher, and in America by Impossible.
We also have a ground beef alternative and chicken burger and nugget alternatives. And ice cream - can't forget the ice cream!
Interestingly, the flavors have to change from one country to another, as every country has its favorites. For example, in Chile we eat a lot of chocolate but in Argentina dulce de leche is a really important flavor so we adapt the product to them. In Brazil we have coconut ice creams because they have a tradition of eating that flavor and everybody loves it.
It's not just the ice cream either, the flavors of our mayonnaise and milk have to adapt too. We've just started operations in Mexico and I'm sure spicy versions of certain products are in the pipeline; those kinds of things simply wouldn't sell in Chile because we aren't big fans of spicy food. You have to look at the market and develop things that suit it, but it's not easy, there is a lot of work behind those changes.
We develop products from the ground up, that's to say that we don't copy recipes or formulas from anywhere else. Everything we create starts from nothing. The products themselves are fairly complex in terms of ingredients and, truthfully, creating the products we created without using Giuseppe would've been incredibly difficult, if not impossible.
One of our slogans, which I love, is ‘expect the unexpected’ because sometimes ingredients appear in our recipes that seem quite strange but if you leave them out then the product starts to lose its quality.
The advantages of artificial intelligence are multiple, firstly, it manages a lot of data without ever making a mistake. Secondly, it has no biases, unlike researchers and developers. We have our biases because we're always considering the plausibility and feasibility of something – it's natural, a part of the scientific method. Unfortunately though, this leads us to eliminate certain possibilities by only considering things we know to be true from previously established knowledge. When using an AI algorithm like Giuseppe, it has the advantage of learning from the vast feedback it receives, as well as by trial and error, constantly making correlations between data that would be impossible to recreate manually.
All this means that you're able to understand the otherwise unimaginable value and function of certain ingredients, and the attributes they can bring to a product. It makes our products original, but also lets us move closer to the qualities of animal products.
A well-designed artificial intelligence using constant feedback, like Giuseppe, improves our speed a lot. It reduces the experimentation process significantly as, instead of starting from scratch, the AI brings you various formulas that have been previously established by trusted learning processes. Through this, you skip one stage of product development. It’s worked really well so far, and has borne fruit in the speed that we are able to release our products.
In general, the products we’ve released recently are superior in terms of their complexity. Our mayonnaise, one of our first products, is much simpler compared with a meat replacement. However, the time it took to create our meat replacements was much shorter than what it took to create Not Mayo or Not Milk, which shows us that the algorithm is bringing us more and more accurate outputs every time we use it.
There is a team of chefs constantly training it, cooking things that may or may not have anything to do with the final product, with the sole purpose of giving the AI the data that it needs in order to cross check it with the data it already has. Then there’s us, the scientists, constantly working from quantitative analysis to translate it into attributes that we hope to have in a product, like its texture profile, 3D structure, or its chemical composition.This helps to make sure that you can get the formula you're looking for more quickly.
Then the next job of the development team is to make sure that everything we learn in the laboratory can be carried over to the industrial process, so that what we make in the lab is what people get on their plates. It's hard work but it’s necessary because, if you tried to copy the formula we use without using NotCo’s technical processes, the product wouldn't be the same.
If you look at our ingredients compared to other products in the market, you'll notice that ours are very different to those of the competition. For example, Not Milk uses pineapple and cabbage and our chicken burger has peaches in it!
As I say, if you start removing those ingredients, you lose the attributes you wanted in the first place. The algorithm is now so accurate that it gives you very specific quantities of each ingredient, and if you stray from the ranges it provides you start to lose the qualities it brings to the food very quickly.
The pulse we use more than any other is pea but we have used lupin beans before. Over time, we started swapping out lupin beans because, although it’s an excellent ingredient, as you move outside of Chile you see more and more people with allergies to lupin, so that made things difficult as we expanded outside our borders into the global market. We also use broad beans and soybeans but the soil depends on the local and geographical circumstances as each country has different levels of acceptability with regards to soy. Countries that produce soy tend to have more interest in eating it.
Why do we use pulses? In the end, they are the best source of plant protein that exists in the world and have healthy interactions with the microbiota of the soils, fixing nitrogen and enriching the soil so that our products have a better concentration of protein. Over the years, one of the biggest criticisms of the vegetarian diet was a lack of high-quality protein, which makes the protein in our base ingredients so important. Pulses dominate in that area of the market.
The biggest gap we have between real meat and meat alternatives is texture, as you say. This is because the proteins that make up animal muscle have a structure that plays an important role solely in the composition of animals, not of plants. Plants don't move, so they don't need muscles and only have cellulose to support them.
So, I can't go into too much detail because the process that we use is confidential, but we use particular instruments that allow us to investigate the attributes of certain ingredients that might have potential for creating better texture. We take the data that we receive from our investigations and feed it into the AI. The AI gives you different hypotheses and we try them out. A team of people make judgments on how close we've come to what we were looking for, which is then measured and feedback is given to the AI. For now, the human part of understanding the sensorial experience of food cannot be reproduced by machines but the machines can provide you quantitative data. We work extremely hard to make sure Giuseppe learns everything it needs, particularly about texture.
There are three important components to texture; chemistry, i.e. its composition; physics, which has to do with the architecture of the food; and finally, the experience, which is everything felt by the consumer, from the sensation in your mouth to how that sensation is then translated by the limbic system in order to invoke the experience in your brain. Often, it has a lot to do with emotions - something that the consumers themselves bring.
I'm not an economist or a businessman but it's certainly significant that such a huge and traditional company sees the way the world is moving towards plant-based foods. I feel like it shows that we have made a difference in changing global food production if a company like Kraft-Heinz, which has such huge volumes, is looking for a younger company like ours that is developing food through tech. It's a step on the way to fulfilling the proposition we made to remove animals from the equation and bring consumers a real alternative.
Disclaimer: The opinions or views expressed in this publication are those of the authors or quoted persons. They do not purport to reflect the opinions or views of the Global Pulse Confederation or its members.
