“If you’re a procurement manager, a procurement analyst in the food supply chain, this is the worst time in your professional career,” says Helios cofounder Francisco Martin-Rayo. “It used to be you would deal with a disruption once every couple of years. Then it became once a year. Now it’s multiple times a year across different regions and different crops. You need help.”
Enter Helios, an AI-powered platform aggregating billions of data points to provide procurement teams and ag commodity traders with a real-time view of the climate and economic risks affecting agricultural commodities from cocoa to strawberries across 90 countries.
The startup, which has just netted a $4.7 million seed round to expand its platform, has been down a couple of blind alleys on the way, but now has a much clearer sense of what the market wants, says Martin-Rayo, a former Boston Consulting Group principal who teamed up with ex-Google AI/ML engineer Eden Canlilar in late 2022 to create Helios.
AgFunderNews (AFN) caught up with Martin-Rayo (FMR) at the World Agri-Tech event in London to get the lowdown.
AFN: What problem is Helios attempting to solve?
FMR: We predict the prices and climate risk of 75 agricultural commodities across 90 countries. We really work for procurement teams. We want to make your life easier so we can tell you when to buy, how much to buy, where to buy from.
We have over 500 billion climate risk data points that are updated every 24 hours. We’ve now ingested all of the USDA reports, we vectorized 40,000 of those reports, plus [reports from] all the other government agencies. We have over 2,000 price series and price forecasts. We scrape 250,000 news sources globally every 15 minutes. We’ve also vectorized that to make it available. So it really is a huge data engineering platform in addition to the work we do.
AFN: How does Helios compare with other AI-powered ag insights startups, some of which have struggled to deliver on their promises?
FMR: There’s a lot of folks that do climate risk well, but they never do any price forecasting. And none of the people who do price forecasting do climate risk. We’re the first company to do both climate risk and price forecasting, which is what we’re hearing the market wants.
You want that single platform that can give you all of the information and all the insights that you need to have a full procurement process, rather than having three, four, or 10 different vendors that you have to go to.
AFN: What commodities are you focusing on and why?
FMR: You have your exchange traded commodities… corn, soy, wheat, etc. But we also have all these other commodities that we call specialty crops… tomatoes, blueberries, raspberries and mangoes that we cover. And so that class is one of the most interesting markets that is typically under reported, under resourced, and yet makes up the majority of the food that we eat.
AFN: In what form and frequency do customers want information so that it’s actionable?
FMR: They want reports. They want someone to chat with. They want an API. Sometimes they want our dashboard. They want it every 24 hours, things are changing so fast. Depending on how large you are and how nimble you are, sometimes you can make decisions relatively quickly and figure out, hey, maybe we should buy more [of commodity X] over the next three months. If you’re a larger Fortune 100 company, maybe you need six months at least before you’re able to make those decisions. So we are able to give you kind of those forecasts 12 months out on the platform.
AFN: How have your customers historically got this information and these insights?
FMR: This was one of the most interesting questions we had when we started the company. We spoke recently to a Fortune 10, the second largest buyer of fruits and vegetables in the world. And we said it probably takes you 24 hours to get on the phone with a Bloomberg analyst… and they said, ‘We never get on the phone.’
And so a lot of the time you are just looking a lot of these qualitative reports, calling your networks, flying to meet your suppliers, and then trying to piece it all together. Before Helios, there wasn’t really that single platform that could give you all of those insights.
AFN: You’ve recently launched a virtual analyst called Helios Horizon – what is that?
FMR: It’s a multi-agent platform. Think of 20 different analysts that we’ve created [using AI]. One does climate risk, one does economic risk, one does news risk, one does geopolitical risk, and so on.
What Helios Horizon does is, you’ll ask it a question like, What is the price of Arabica coffee [the main futures contract traded in New York] going to be in six months? What is the price of mangoes going to be? What is the climate risk going to be in Indonesia for pineapples?
And it’ll look at the question and figure out what the different steps are, and then in a couple of minutes, it’ll give you an answer that directly gives you what is the risk? When should you buy? How much should you buy? These are tasks and analyses that would have taken weeks before, and now it takes you a couple of minutes. You use different AI agents depending on the question you’re asking.
AFN: How have you validated your platform?
FMR: For prices, we’re five to 20 times more accurate than the industry standard. We did it [validated the models] the traditional data science way. So you have 10 years of data, you train your model on nine years of data, and then you test it on one year. We rebuilt a lot of the standard industry models, and then we compare them to our own and how close we are to forecasting that price, what we call the error rate. That’s what drives our accuracy ratings.
And then for climate risk, we’ve identified 90% of all the disruptions that have happened for all 75 of the agricultural commodities we cover, which is unprecedented. Again, we look at it from a back testing perspective and say, All right, well, here are all the disruptions that have occurred. How many did our platform cover?
We saved Libby’s, the second largest buyer of mandarins and pineapples in the United States, 15% on their cost of goods sold for mandarins because we had predicted a pretty significant increase in prices months before the market did.
So they were able to buy more and buy earlier than they would have otherwise, getting that extra 15% on their bottom line, because you’re pulling all these different data points and sources in a way that gives you the insight that is, hey, we think the price of mandarins coming from China is going to increase pretty dramatically in six months. And here’s why.
And so it’s always giving you that transparency about here’s why we think that this is going to happen, and then you as a procurement specialist are able to use that information as you see fit.
I think one of the more interesting questions we talked to with larger organizations is we’ve spent tens or hundreds of millions of dollars of CapEx building out these logistics facilities, transportation facilities, processing facilities. Am I going to be able to get my barley from the same place five and 10 years from now? If not, where should I start to build out that infrastructure and those relationships? That’s the core of our climate and supplier audit offerings.
But what’s exciting about Helios Horizon is that you can just go on the platform and say, ‘Am I still going to be able to get barley from this particular area in Brazil or the United States, five or 10 years from now, what’s the risk look like?’ And you can get an answer back in a couple of minutes.
AFN: How has the platform evolved over time? Did you make any early mistakes or go down blind alleys?
FMR: When we first started, we had a very large enterprise customer, and their view of the market was very supplier specific. They said, Can you give me the individual risk for all of my suppliers? So we built our initial MVP around that.
We took it to market and what people said is, I know my suppliers really well. What I don’t know is the rest of the market, so can you build out the platform and tell me what the rest of the world is doing?
And that was a huge pivot for us, where it became much less about your supply chain period, rather your supply chain and the broader global context.
AFN: How have you funded the platform?
FMR: We’ve had a lot of support from the VC community. We raised our $3 million pre seed from Supply Change Capital as the lead, and we’ve just announced our seed raise [$4.7 million] led by Collide Capital and S&P Global Ventures.
This for us, is a really nice market indicator that what we’ve built is the way of the future. This is how people are going to engage with virtual analysts and get the insights that they need, custom insights at scale, which has never happened before.
AFN: Can you name any high-profile clients?
FMR: We were very lucky to win Walmart’s open call last year and they’ve been a wonderful partner and customer for us. We have a lot of other big contracts, but we can’t typically talk about our customer names.
The post Helios taps AI agents for ag insights at ‘worst time’ for procurement pros appeared first on AgFunderNews.
