He didn’t grow up on a farm, but David Ahmed was still born with green fingers. By the time he got to university in New York, he was spending hours in his dorm room tinkering with lettuce plants to understand how they were affected by different light spectrums.
As a computer scientist, Ahmed already had a passion for optimisation. He just wasn’t sure exactly what to optimise. It was his lettuce plants that eventually provided the clue.
“I was building these little farms with sensors, changing the variables, and getting some very odd results,” he says. “It got me thinking: can we run thousands of these experiments, let computers play with the variables, and basically find the holy grail of agriculture?”
Ahmed came to believe that even the best growers were massively underutilising the potential data available to them. “What I was thinking, and it turned out to be true, was there must be lots of growers who are growing, say, tomatoes, but what are they doing to really understand it? They’re basically doing nothing.”
Without the funding to buy his own farm, Ahmed instead built a software programme for vertical farms that he thought could be used to make sense of all the available data and help any grower maximise production. He envisioned it as “the AWS [Amazon Web Services] of ag”.
“It was basically: you bring the farm, we’ll bring the intelligence,” he explains. “So if you’re putting tomato seeds in your greenhouse and want the maximum yield in the third week of November, the factors to play with are impossible for any human being to tune. We think our models actually understand those factors and can do it much better.”
From dorm-room lettuce to a €4.8m fundraise
Fast forward to today and Ahmed’s idea is now Berlin-based start-up Hexafarms. It is no longer just the AWS dream, but a full operating system combining advanced sensors, AI and robots, used by more than 50 growers across 13 countries.
Customers include SanLucar and grower cooperatives Elo and Royal ZON. Hexafarms says customers have managed more than €300 million worth of produce through its technology this year, while its AI models currently cover around 50 cultivars.
Over the last five years, Hexafarms has built a suite of technologies it says can monitor everything from carbon dioxide and temperature to humidity and solar radiation, feeding the data into AI models designed to help growers achieve the biggest possible harvest at the right time.
Investors are impressed too. Hexafarms has just announced a €4.8 million seed round to develop its wireless crop sensing and automated pest and disease scouting technology and accelerate expansion across DACH, Benelux and Spain.
The round was led by Ananda Impact Ventures and Green Generation Fund, with participation from Grey Silo Ventures and better ventures, alongside existing backers Speedinvest, Mudcake and Techstars. The company has also secured Rentenbank’s Startup Grant and a subordinated loan of up to €800,000.
Can growers really be leaving 30% on the table?
“We want to become the window into commercial food production,” Ahmed says. “And we bring everything: hardware, software, AI machine learning model, all the SaaS modules needed. Basically, one place you get it all.
“We strongly believe that even the best of the growers are leaving 30% on the table in revenue improvement,” he continues. “And our claim is that we will be able to give that back to you, driven by what our machine learning and AI models learn in different optimisation strategies.”
It is a computer scientist’s solution to a problem many farmers will feel they have been tackling for generations: plant, grow, harvest, learn, then try something different next time. It’s a method as old as agriculture itself.
Hexafarms, by contrast, is still younger than a single crop rotation. Given it relies on AI models that are only ever as smart as the information they’re fed, how can Ahmed be confident Hexafarms has enough data, and has seen enough harvests, for its models to predict the best growing strategies?
‘We have more data than Bayer and Syngenta’
“I’m going to make a claim that sometimes, actually most of the time, we have more data on the crop than the seed breeders themselves,” he says. “We have more data than Bayer and Syngenta would have on their own crops.”
The reason, he argues, is that Bayer or Syngenta might conduct numerous longitudinal studies involving thousands of experiments, but Hexafarms is effectively doing that on every farm, every season, analysing thousands of tomato plants each day for their fruit and flower count, stem diameter and leaf size, alongside data on what is going into the crop and what is coming out.
“A lot of these [seed breeder] labs would go and study the plants, and they’ll write down we had four tomatoes. But where’s the picture of the tomatoes? How big were the tomatoes? That’s the data that we have that other companies don’t.”
Ahmed says eight out of 10 human experts surveyed by the company now say they trust Hexafarms more than their own intuition.
‘We are literally building three or four companies’
Ahmed emphasises, however, that none of this has come easily. Hexafarms has developed its own low-cost, high-resolution cameras capable of transmitting data from almost anywhere in the world. It also has wireless sensors “that are like lab-grade instruments” while remaining relatively inexpensive. Then there is the AI model itself.
“We are literally building three or four companies under one company,” he says. “That’s the difficult part here.”
Now, Hexafarms is turning its attention to robotics, based on the belief that improving its hardware will deliver the next big leap forward.
“We started with four camera lenses looking per hectare. Now we are launching robots which will be covering every single plant. So our machine learning models are going to be bombarded with more and more data,” he says.
“We are nowhere close to the plateau that we will eventually hit in this optimisation game.”
