April 7, 2022
![Artificial intelligence discovers ingredients for food, supplements Artificial intelligence discovers ingredients for food, supplements](https://eu-images.contentstack.com/v3/assets/blt09e5e63517a16184/blt3bb71d14f84952df/64ff18795d45a2879d16e298/PROMO-Technology_20Trends-838379014_0.jpg?width=700&auto=webp&quality=80&disable=upscale)
The tech: NemaStudio.ai and the Infinity Chip What it does: Analyzes how compounds impact the lifespan of tiny worms living in a microfluid environment. Founded: 2019 Like many of these biotech startups, NemaLife began in an academic lab (Texas Tech). Except this origin story doesn't just involve computers. Worms are at play, too. C. elegans are roundworms with a genome strikingly similar to humans'. About 40% of their genes are comparable to those found in people, which makes them extremely handy models for human conditions. Fans of these tiny worms, like Patel, say they can be more convenient test subjects than rodents because they are transparent, live their entire lifespan in just two or three weeks, and take up much less space than mice. At 1 millimeter long, roughly 70 animals can live on each of NemaLife's patented 1 inch x 2 inches microfluidic chips, where they can be observed as they are fed various test compounds. Using worms to model human conditions isn't new. What's novel is the microfluidic platform and AI data analysis engine NemaLife created to record, analyze and interpret the worms' every move while they're being fed test compounds. "Let's say you're recording 500 to 1,000 videos (of worms) each day. To have a person watch every single one of those videos and manually record all of those events, it just wouldn't be feasible," Patel explains. The AI platform, named NemaStudio, has been trained to identify worms and measure how much they are moving to indicate health and death. One of NemaLife's first commercial clients was Canopy Growth, a cannabis company that scored a surprising discovery thanks to the worms. "Canopy Growth initially just wanted safety data on the effects of chronic CBD use," Patel says. "And we found that a low dose CBD actually extended the worm's lifespan by at least 10%." Of course, this isn't where testing ends. Worms, after all, aren't humans. "But the AI gets you actionable data at lower costs, really helping you make smarter investment choices for more complicated experiments later down the R&D pipeline," Patel says.
The tech: Giuseppe What it does: Discovers new combinations of plant-based ingredients to replace animal-based foods. Founded: 2015 NotCo is an alt-protein startup that achieved unicorn status in 2021, a year flush with funding from star investors such as tennis legend Roger Federer, and partnerships with Burger King in Latin America and food giant Kraft Heinz. Giuseppe gets most of the credit. Giuseppe is NotCo's AI platform, which leverages machine learning algorithms to find the best plant-based replacements for animal proteins by considering everything from flavor to viscosity. Giuseppe's discoveries have led to multiple products—NotMilk, NotBurger, NotIceCream, NotChicken and NotMayo— that are sold in the U.S., Canada, Brazil, Argentina, Chile, Mexico, Peru and Colombia. Karim Pichara, co-founder and CTO of the Chilean food tech firm, says Giuseppe generates concepts, recipes and processes that are tested in NotCo's kitchen. Critiques are fed back into Giuseppe so the AI can continue evolving. Giuseppe has come up with some seemingly random yet highly successful concepts, he says. The biggest surprise to date is the recipe for NotMilk, which includes pineapple and cabbage. It turns out that pineapples share molecular similarities to cow milk. NotCo introduced NotMilk to the U.S. in 2020 and debuted NotBurger—made with cocoa and spinach powders—at Natural Products Expo West in March. The NotChicken Burger, whose ingredients include peaches, is coming to the U.S. later this year. "Only a computer system can, without any bias, analyze information, learn and propose things that are totally different compared to what has been done in the past," Pichara says.
The tech: Forager What it does: Builds a massive library of plant compounds and their connections to human health. Founded: 2017 Brightseed is an AI-powered data company cataloging millions of hidden compounds in plants and illuminating the medicinal and nutritional dark matter of the plant kingdom. The San Francisco, California-based startup's technology, named Forager, uses machine learning that's been trained on a massive library of biomedical and plant research. Forager can answer two types of client questions: What beneficial compounds are hidden in their plant sources? Which plant sources should they include in their products to deliver specific health benefits? "Today we've mapped over 1.5 million compounds to human health targets, with a goal of mapping 10 million by 2025," says Jim Flatt, Ph.D., Brightseed co-founder and CEO. It's an impressive start that has attracted global attention and partnerships. Brightseed saw a 300% increase in customer growth by the end of 2021, with clients coming from agricultural, food and beverage, dietary supplement, nutritional and personal care spaces, according to Flatt. Notable public partnerships include Danone, Pharmavite, Ocean Spray and the Bill and Melinda Gates Foundation. Ocean Spray enlisted Forager to create a comprehensive nutritional profile of its cranberry varietals. Forager's analysis revealed 10 times more phytochemicals and four times more bioactives than previously known. According to Ocean Spray's 2021 press release, Forager also found that multiple cranberry strains contained more than 350 bioactive compound classes with potential impacts on immunity and cognition—health benefits not previously linked to cranberries. Brightseed also discovers, validates and commercializes its own bioactives. In its first major discovery, Forager identified N-trans caffeoyltyramine (NCT) and N-trans-feruloyltyramine (NFT) in more than 80 common edible plant sources, including black pepper and hemp seed shells. Preclinical research studies established NCT and NFT as metabolic regulators with implications for fatty liver disease, diabetes, cardiovascular disease prevention and treatment. Brightseed, which received a 2020 NBJ Award for Science & Innovation and was named a 2022 SXSW Innovation Award finalist, has future plans to expand beyond edible and medicinal plants to map fungi and beneficial bacteria used in fermented foods and probiotics.
The tech: #foodbrain What it does: Crawls more than 28,900 online data sources to provide insights for food and beverage companies Launched: 2019 Spoonshot grew out of a lesser-known consumer mobile app called DishQ, designed to provide APIs for restaurants and make customized food suggestions for diners. While in beta, the app attracted the attention of chocolate giant Ferrero, which invited DishQ co-founders Kishan Vasani and Sai Sreenivas to speak to Ferrero's team about the future of AI. In preparation, Vasani says he and Sreenivas got creative. They trained their sights on Ferrero's popular Nutella product and sent their AI crawling through diverse online data. The results were surprising: The AI discovered that Nutella's dominant ingredients have an 80% flavor compound match with soy sauce. It also found that Asian consumers were chatting online about Nutella/soy sauce noodle recipes. "In our presentation, when I said, 'Nutella soy sauce—it works,' there was silence in the room. I was sitting there a bit nervous," Vasani says. "And then they started laughing and asking questions about how we got our conclusion." "That was really the lightbulb moment for me," Vasani continues. "If Ferrero hadn't thought about Nutella in this way, there must be a ton of insights and data out there we could use to educate brands." Vasani and Sreenivas went on to develop Spoonshot, powered by AI named #foodbrain, which crawls 28,900 public data sources. It looks for ingredients, products, consumer reviews, research papers, menus, recipes, articles and social conversations. Then #foodbrain structures the data and "connects the data dots" using food science. Spoonshot, which works as a self-service insight platform and as a paid subscription for deeper insights, is attracting growing interest. It raised $1.8 million by 2020, and Vasani says revenues tripled in 2021. "This is the future. This is how you do cost effective, agile, high velocity, high volume research on a big scale," Vasani says.
The tech: Nπϕ (Nuritas Peptide Finder) What it does: Analyzes hidden peptides (small proteins) in plants and natural food sources to predict and identify how they impact health Founded: 2014 "When you are discovering a whole new angle to a plant material that, as humans, we've never really used—maybe it's anti-inflammatory, reduces muscle atrophy, improves sleep—I think that's mind-blowing," says Nora Khaldi, founder and CEO of Nuritas. Nuritas is a Dublin, Ireland-based biotech company that combines AI and molecular-level 'omic anlysis to build the world's largest peptide knowledge base. The Nuritas platform, named Nπϕ (Nuritas Peptide Finder), analyzes billions of peptides in plants and natural food sources to predict and identify how they impact specific health areas, molecular pathways or receptors. After Nπϕ makes its predictions, the Nuritas scientists produce the peptide networks and perform cell-based tests. In a short time, Nuritas has discovered thousands of active peptides, filed for 65 patents, partnered with the European Union on a diabetes prevention study, and, most recently, closed a $45 million Series B round, bringing its total funding to $75 million. Early investors included U2's Bono and The Edge and Salesforce CEO Marc Benioff. Partners and customers have included Nestlé, Mars, BASF and Pharmavite. The work led to several branded ingredients: PeptAIde 4, a peptide found in rice with potential benefits for hair and skin damaged by inflammation; PeptiYouth, discovered from peas and supporting skin health; and PeptiStrong, discovered in fava beans and supporting muscle health. "I think the most important thing I learned here is when you eat the source food, you don't necessarily get those benefits," Khaldi says. "Your body does not assimilate them. You actually break them down. What we're doing with the AI is identifying the peptides and making them bioavailable by concentrating them."
The tech: Flourish What it does: Identifies ingredients that replicate the taste, texture, gelation, foaming, emulsification and binding abilities of animal proteins. Founded: 2019 Shiru, a San Francisco, California-based startup, embarked on an environmental mission to reduce our reliance on livestock by redesigning foods from novel ingredients from plant-based proteins. The challenge is finding proteins that taste, cook, look and chew like meat, dairy and eggs. Shiru believes its AI platform Flourish can find the right recipe. Flourish maps protein identities to food functionality by using computational biology to interrogate a massive database of millions of naturally occurring proteins, then tests any promising candidates to evaluate their compatibility with the desired ingredient functionality. Without Flourish's help, finding the ideal candidate for, say, a non-egg protein that can act as a food binder, would be like searching for a needle in a haystack, explains Julian Lewis, Shiru's vice president of business development. "There are around 50,000 known edible plants on Earth. And every single plant has about 30,000 to 40,000 different proteins," Lewis says. "You can do the math…There's a huge amount of data that's not connected to food functionality. Flourish helps us access it." Once Flourish identifies a novel plant protein, Shiru doesn't go to the ends of the Earth to harvest it. (It's probably not sustainable or economical.) Instead, the food scientists use small samples to genetically engineer and produce the proteins in large quantities via microbial fermentation—also known as precision fermentation, something we're going to hear a lot more about in the future. "Precision fermentation is being recognized as a fantastic sustainable way to produce materials, foods and proteins of the future," Lewis says. "It's the frontier of tech. There is going to be an explosion in the future."
The tech: Flourish What it does: Identifies ingredients that replicate the taste, texture, gelation, foaming, emulsification and binding abilities of animal proteins. Founded: 2019 Shiru, a San Francisco, California-based startup, embarked on an environmental mission to reduce our reliance on livestock by redesigning foods from novel ingredients from plant-based proteins. The challenge is finding proteins that taste, cook, look and chew like meat, dairy and eggs. Shiru believes its AI platform Flourish can find the right recipe. Flourish maps protein identities to food functionality by using computational biology to interrogate a massive database of millions of naturally occurring proteins, then tests any promising candidates to evaluate their compatibility with the desired ingredient functionality. Without Flourish's help, finding the ideal candidate for, say, a non-egg protein that can act as a food binder, would be like searching for a needle in a haystack, explains Julian Lewis, Shiru's vice president of business development. "There are around 50,000 known edible plants on Earth. And every single plant has about 30,000 to 40,000 different proteins," Lewis says. "You can do the math…There's a huge amount of data that's not connected to food functionality. Flourish helps us access it." Once Flourish identifies a novel plant protein, Shiru doesn't go to the ends of the Earth to harvest it. (It's probably not sustainable or economical.) Instead, the food scientists use small samples to genetically engineer and produce the proteins in large quantities via microbial fermentation—also known as precision fermentation, something we're going to hear a lot more about in the future. "Precision fermentation is being recognized as a fantastic sustainable way to produce materials, foods and proteins of the future," Lewis says. "It's the frontier of tech. There is going to be an explosion in the future."
The next new ingredient isn't being discovered on a farm or in a kitchen. It's coming from deep inside a computer that's always learning and evolving.
This isn't science fiction. It's the future of R&D.
A new generation of computer experts is pairing food science and molecular biology with powerful artificial intelligence to uncover natural ingredients, formulations and trends that we slow-brained humans never could have imagined.
Like, who would have thought that pineapple shared molecular similarities to cow milk and could be a key ingredient in a non-dairy milk beverage? Or black pepper might hold the key to liver disease prevention and treatment? Or soy sauce and Nutella make a fantastic flavor combination on noodles?
AI did.
Big-data technologies like natural language processing, computer vision and deep learning are being used to reverse-engineer ingredients from the molecule up and at the computational speed of light.
"It's a massive step for humankind because this is the first time we have AI-discovered ingredients from nature. They are not designed by artificial intelligence; they are discovered by it," says Nora Khaldi, founder and CEO of Nuritas, which uses AI to look for new, bioactive peptides in plants and familiar food sources.
The AI trend has been growing within the food, beverage and nutraceuticals industries for about five years, as competitors race to find the next plant-based proteins, supercultures for fermentation and bioactives hidden in the plant kingdom. Biotech and computational biology startups are proliferating, and a few, like NotCo, have reached unicorn status because of this hunger for speedy innovation.
"You have the ability to move new products to market really, really quickly if you're using AI," says Dhaval Patel, Ph.D., director of research and innovation for NemaLife, a TechBio company that uses AI microfluidics and tiny worms to study the health benefits of new compounds.
Click through this gallery to meet some players in this emerging space.
About the Author(s)
You May Also Like