Colombian entrepreneur Felipe Ayerbe wants to turn the everyday morning cup of coffee upside down. With a civil engineering degree from Universidad de Los Andes in Bogotá, the former consultant at Accenture and McKinsey is also experienced in investment banks and asset-management firms including Corficolombiana, Credit Suisse, and Sanford Management. But last year he decided to start a company to work with tropical agriculture and co-founded Demetria with the current executive chairman Eduardo Shoval, CTO Yori Nelken, and exec Salomon Kassin.
Demetria is an agritech that developed an artificial-intelligence platform to determine the taste and quality of green coffee beans. “Like a lot of good ideas, Demetria arose through luck, but because we were looking for luck. The whole idea started when we went on a trip to Israel to try finding technology that would be applicable to tropical crops. We thought there was a vacuum, a lack of services of technology, and Israel evidently is one of the best innovating countries in producing technology for the agricultural space, but it’s typically done for the non-tropical crops,” Ayerbe, the CEO, told LABS in an interview.
Ayerbe knew about the 40-year-old hardware sensors’ technology early on in his career as an engineer, but at that time, machines were so big that they couldn’t fit in a room. Yet, visiting Israel, he reached out to a company that produced handheld sensors and thought about applying them to the $450 billion technology-less coffee industry.
“The coffee industry has pretty much had the same thing going on for 200 years,” he said.
Demetria uses the sensor to read what it calls “the spectrum fingerprint” of the bean, which is a proxy for the whole chemical composition. The problem is that it is not in a language that people can understand; instead, it’s a spectrum language.
Historically, the industry has translated that spectrum fingerprint into chemical composition, using this technology to determine the amount of sugar in sugar cane or how much fat protein there is in milk, for instance.
But Demetria’s “eureka moment” was when it correlated the spectrum fingerprint to taste, rather than to chemistry. “At the end of the day, the taste is a representation of chemistry, but it’s the representation that we have as humans,” said Ayerbe.
With that, Demetria could replace the time-consuming manual process of cupping coffee by automating the analysis of the green coffee beans at any point during the process. How? By using these sensors and its software, i.e, the AI platform to better understand how the coffee will taste for consumers. Before this, coffee farmers, traders, and roasters only knew about the taste by the artisanal cupping, usually far down the supply-chain journey.
“We were able to prove we can measure taste in green beans before roasting. This is something that, when we started the process, a lot of people in the industry thought we were crazy, but last November we were able to prove not only that it could be done, but it could be done with a very high degree of certainty. We are over 90% accurate, which is comparable to what an individual cupper is,”, he said.
Cuppers are people who taste coffee. As people, they have different tastes, allergies, and all those human kinds of things. But, even though Demetria can replicate cuppers with its technology, Ayerbe says his objective is not to replace human workers, but to add technology to the process.
“The problem with the industry is that because cupping is so difficult to do, and it’s so exclusive, it’s not good enough. So the most important variable in the industry, which is quality and taste, is not measured enough, and that’s why the vast majority of farmers in the world have never tasted their coffee; they don’t know what they’re producing. Consequently, they don’t know what they need to do in the field to produce better-quality coffee, because what you don’t measure, you don’t manage.”
By automating the process, Dementia’s technology increases the reproduction of high-value coffee seedlings to provide high-quality coffee plants, pushing efficiency and transparency throughout the commodity’s value chain for the 12.5 million small coffee farmers. Yet, Demetria is not only targeting the small ones.
The newcomer has already caught the attention of one of the world’s top roasters Nespresso, besides agricultural commodities merchant Volcafe/ED&F Man and the Colombian Coffee Growers Federation.
Making specialty coffee, well, special
In an industry where taste is king, Demetriais already generating revenues. Nespresso develops coffee varieties through the grafting of coffee-tree seedlings, a horticultural technique used to join two plants so they grow as a single plant. To date, Nespresso relied on the experience of a few trained experts to manually examine the seedling to detect if the grafting process was successful and the plant was viable.
Now, Demetria’s SaaS platform reads the coffee bean using near-infra-red sensor technology and AI-driven data intelligence to enable Nespresso’s operational teams to measure and classify grafted stems of specific coffee-tree seedlings.
Before the commercial agreement, Nespresso underwent a pilot with Demetria’s app to classify more than 240,000 seedlings in a three-month period that have since been supplied to Nespresso’s select network of farmers in Colombia.
Nespresso’s deal is important because Demetria’s main performance indicator is contracted value. As with everything in a tech-related business, it’s all about the data. Because once the firm engages with a customer, the signed contract gives Demetria the data to codify their cupping-quality experience and deliver applications.
Ayerbe believes that by the end of this year, the company will have about $4 million in total contract value. “Revenues are going to happen a little bit with a lag because first we get the data and then we deliver the applications. This year, it will be a little less than $1 million [in revenues],” he explained.
“Our concentration is to work with the largest players in the world. The reason being is because the coffee market has millions of farmers but relatively few traders and roasters that manage a big percentage of the coffee flow. So by concentrating on those, we are able to codify the data that we can take worldwide,” stated the CEO.
“Once we do that, we are going to go to the farmers because this is a top-down perspective. If you want to create a new standard, you start with the people who value quality today, which are the coffee buyers. The farmers do now know the quality. So once we do that, we go to the farmers, and then deploy the application for the farmers, and we can have the whole value chain connected.”
Ayerbe said the coffee industry is going through a “winefication” process, and Demetria is enabling the digitalization of that. Meaning that the farmer can see through the cloud the exact type of buyer looking for gourmet coffee.
“Forty years ago, you would ask for coffee and you would get it for free with a very low quality. It was just a method to deliver caffeine. Fast-forward to today when you go to a fancy restaurant and you ask for coffee when you get a menu. You have different choices, different price points, different preparations, and that’s what we call the winefication of coffee,” he said.
The winery is a sophisticated and profitable business. But coffee’s money has so far been concentrated in commodities. For a farmer, the value difference between the mainstream-commodity coffee and the high-value specialty coffee can be 30 times, and even reach 90 times, according to Ayerbe.
“If a farmer can manage his quality and measure and understand the quality, having transparency to understand who is going to be able to buy that cup of coffee at the best price possible can be a difference of 30% and 40% in the value of revenues for that farmer. Whenever you place transparency in the business, the industry grows,” he said.
Not a LatAm agtech, but a worldwide company
Although the startup is a very young company, it already has a relatively global footprint. Demetria’s development team is based in Israel, and its “on-the-ground” team in LatAm is based in Colombia. The startup opened up an operation in Brazil about three months ago, setting up a lab in Varginha, the hub of the coffee region in Brazil. It also has a commercial presence in Europe.
Born amid the pandemic, the 35-employee team was never tightened down to physical constraints. “We feel that we are going to be able to keep on growing with that global footprint, and hiring the best talent no matter where they are. Yet, the commercial side and the management side have to be where the coffee is, so that’s why we are in Colombia and Brazil, and probably soon in Vietnam.”
Those three countries produce about 60% of the world’s coffee. So with a relatively “small” presence, Demetria can cover the biggest amount of coffee production in the world. Demetria’s technology works with high-value, specialty coffee as well the mainstream ones. Colombia has been distinguished for mostly producing high-value coffees, and Brazil has a little bit of both, robust high-value coffees and more generic commodity coffees.
Demetria has already raised a $3 million funding round led by LatAm-Israeli investor Celeritas and a group of private investors including Mercantil Colpatria, the investment branch of Grupo Colpatria, a leading player in the Colombian financial sector, and it’s on its way to raising a Series A round to ramp up the AI and the delivery-service team.
The early-stage firm plans to launch a suite of SaaS-based solutions to deliver taste assessment and profiling of green coffee beans as well as swift, accurate quality measurement and traceability throughout the supply chain.
Traceability is a big issue, not only in the coffee industry, but for food in general. Demetria found that the spectrum fingerprint of a particular batch of coffee is unique enough to use that like a human fingerprint. It can identify that the same batch of coffee from a farm in Brazil is the same one received at another port somewhere around the globe, for example.
“People want to have a connection to their food and they want to understand where it is coming from; not only from a fraud-prevention perspective but also from the brand story. In most of the solutions in traceability, you can trace the packaging, the transactions. But here, we’re literally tracing the physical bean,” said Ayerbe.
“While the money flows in traditional innovation hubs like Israel and Europe focused on solving the problems of first-world agriculture, there is still a market to expand and innovate for the third-world’s tropical crops.”
“Our Latin American countries are still very dependent on agriculture. The impact we believe we can bring with agtech is much more important here than in other areas, so I believe that this is something that hopefully more people like us can solve,” he added.