AI-powered startup Helios has launched what it claims is the world’s first global platform to forecast prices and availability of soft commodities from fruits and vegetables to cocoa.
Founded in December 2022 by ex-Google AI/ML engineer Eden Canlilar (CTO) and former Boston Consulting Group principal Francisco Martin-Rayo (CEO), Virginia-based Helios uses AI to translate climate, economic, currency and political risk signals into actionable insights for CPG (consumer packaged goods) companies, hedge fund managers, and agricultural commodity traders.
Helios has a granular view of how climate affects a crop’s growing season globally, with daily and historical climate data for over 14 million locations, claims Martin-Rayo, who says interested parties can sign up for a free trial and onboard themselves in minutes.
“For the first time… users can monitor a crop’s growing season across key countries, quickly determine supply availability and forecast global price shifts. You choose the commodities and growing regions that matter most to you and receive custom weekly reports that highlight major risks, trends, and changes.”
Strategic pivot: ‘It turns out that customers didn’t want conversations with a chatbot…’
Helios launched in beta mode last spring with a platform focused on anticipating risks in customers’ own supply chains by asking them to input their key suppliers and growing regions. This was followed in the fall by a ChatGPT-inspired virtual supply chain assistant called Cersi.
“So it was a case of, ‘Let us know where your suppliers are, and we can tell you which have the highest risk of being disrupted,’” Martin-Rayo told AgFunderNews. “But as we onboarded people onto the platform, they started telling us, ‘This is really helpful, but can you tell me how the rest of the world is doing.’”
As for the Cersi chatbot, he said, “We started to see that people kept asking Cersi the same questions over and over. What customers really wanted was not conversations with a chatbot, but timely customized reports in their email that they can take to their meetings. The beauty of Cersi is that we use her to generate these custom reports for each commodity, so she’s powering the back end, rather than being customer-facing now.”
He added: “I think one of the mistakes some AI startups make is that they get so excited about how cool the technology is, without listening to what customers actually want. And what customers were telling us is, ‘We’re actually pretty good at understanding the risks in our own supply chains [with existing suppliers]. What we don’t have a good handle on is how the global green bean or okra or potato growing season is going.’”
Cocoa, potatoes, mangos…
In the case of mangos, said Martin-Rayo, “Customers were telling us, ‘Don’t tell me just about my mango suppliers in Peru, but tell me how the mango season in Vietnam, Mexico and India is going.’
“For the first time, [users of Helios] can see how the global growing season is going for 50+ commodities. If you’re buying corn, you just buy corn futures. But if you’re trying to buy okra, which is not exchange traded, you don’t know how the global commodity is doing and you’re highly dependent on word of mouth, calling suppliers, and reading newsletters that are coming out every now and then. But there isn’t a way for you to say at this point in time, how is the global okra season going and how might prices change?
“With Helios, you can see how any particular commodity that isn’t exchange traded is doing and we use risks signals to come up with ratings. Over the past few months, we’ve been using these ratings with hedge funds to help them predict where the price of commodities is going. We did it for cocoa and it was incredibly successful. We predicted the increase in prices months before the market.”
“Helios has unmatched climate risk signals in the industry that are highly predictive of soft commodity prices.” Daniel Goldberg, Coresight Research
For cocoa, in early to mid-December last year, Helios “saw the risk rating really shoot up to a vastly different level than it has been historically and then at the end of January, it got a lot worse, and this was incredibly predictive,” said Martin-Rayo. “So we could say to our customers, ‘Hey, cocoa is looking a lot riskier than it has been and we’re seeing a lot of climate issues.’ The hedge funds added these signals into existing models to give them a buy or sell signal.”
He added: “What’s really valuable about our platform is that we have 20 years of climate data. And so for cocoa, we could see the average risk was ticking up in 2020, 2021, and 2022, but in late 2023 and 2024, it really started to increase dramatically. Using our platform you can see not just a prediction of where the price might be going next month, but what the trend looking like longer-range.”
The same applies to palm oil, he said. “We started to see the same levels of increased average risk and increased volatility, so we expect to continue to see prices be squeezed.”
Helios’ platform was also able to predict challenges in the potato crop in parts of Europe well ahead of time, he said noting that growers in France, the UK, and Belgium lost a meaningful percentage of their potato crop last year. “Helios started flagging a few weeks before the market that there was an enormous amount of excess precipitation that you don’t typically see, which meant they weren’t able to harvest all of the crop, while some was also lost in storage.”
Risk scores in real time
But how does the information Helios can provide compare to what buyers or traders can already see using existing tools?
Right now, said Martin-Rayo, “Hedge funds are waiting for USDA reports or a Brazilian report on corn that comes out in Portuguese that they throw into Google translate, and other delayed sources of information. We have the ability to get risk scores in real time based on custom AI/ML models specific to a particular crop.”
He added: “Again, right now, hedge fund managers can call meteorologists who can say, hey, there’s a drought in Brazil or El Nino effects in Peru [which may impact pricing and supply]. But that’s only part of the picture. Take mangos in Peru, where there’s been an abysmal harvest. Everybody was focused on precipitation, but in fact we could see from our platform that the rain was actually perfectly in line with historical patterns. The real problem was temperature; the highs were too warm during flowering time [too much heat can lead to less flowering and lower volumes for harvesting once fruit forms].”
“I was at a conference the other day and we were talking to a third-generation grower in India and he said: ‘My father and grandfather would expect a problem every now and then [due to extreme weather]. Now, every year, one of the crops is going to fail.” Francisco Martin-Rayo, CEO, Helios
‘VCs would ask us, What software systems are your customers currently using? And I’d say WhatsApp and Excel…’
Pricing for Helios depends on the type and number of commodities users are interested in, said Martin-Rayo. “For [custom reports on] exchange-traded commodities such as cocoa, corn, soybeans, palm oil, and cotton, it will cost you $995 a month.
“But for everything else, from tomatoes and potatoes to okra, spinach, blueberries, blackberries, onions, and so on, it’s from $399 a month. We want the information to be accessible and affordable and democratize access to these AI-based insights,” said Martin-Rayo.
He added: “We raised a pre-seed round last year [led by Supply Change Capital with participation from January Ventures] and VCs would ask us, ‘What software systems are your customers currently using? And I’d say WhatsApp and Excel.’ There hasn’t been a great partnership with technology because they haven’t seen any real value from it, yet. If AI-based platforms like ours are to succeed, we have to provide actionable information that is going to help with business decisions.”
Better price forecasting for commodities that are not exchange traded
So how does Helios compare with other AI-powered ag insights startups, some of which have struggled to deliver on their promises?
“One of the mistakes I think that companies in the space have made is to get really excited about predicting yield and focusing on a handful of commodities [such as soybeans],” claimed Martin-Rayo.
“We’ve really anchored on our ability to do better price forecasting for a lot of these commodities that people are ignoring. Yields matter, but what really matters to our customers is price. You look at some of the companies in the space and say, I can predict where yield is going to be two or three months from now. But price can be totally uncorrelated with that.”
Where is the data coming from?
Asked what data Helios is scraping to provide users with actionable insights, Martin-Rayo explained: “We’re getting satellite and weather station information for climate data, things like daily highs and lows average temperature, cloud cover, humidity, precipitation. For economic risks, a big focus has been on currency volatility, energy prices and then a basket of fertilizers.
“And for catastrophic risks we built this incredible NLP [natural language processing] engine that looks at half a million news articles across 60,000 global news sources every day. Every 15 minutes we’re tracking force majeure events,” he added.
“Right now you might be getting a lot of data, but it’s just being dumped on your supply chain analyst; we’re turning it into something useful.”
Guest article: The real impact of tariffs on US produce prices