Vivid Machines, FruitCast Push AI Into Harvest Timing Decisions

Last year's apple harvest at Okanagan Specialty Fruits in Washington State started on schedule, then stopped cold by mid-morning. "It was like 38C … it's not safe for people to work in that heat," says Joel Carter, who works at the company. "We had to stop at 10 o'clock in the morning."

That kind of disruption is the problem a small group of agricultural technology firms is trying to solve with artificial intelligence. New software and camera systems now forecast not just when fruit will ripen, but how long growers have to pick it before heat, disease or market prices turn against them, according to the BBC.

What problem is the AI solving for growers?

Timing a harvest wrong carries a direct financial cost. If a grower books seasonal pickers before fruit is ready, the labor goes to waste. If pickers arrive too late, the crop can spoil or miss a price window — a particular risk for high-value, fast-ripening fruit such as strawberries and blueberries, whose prices can swing sharply from week to week.

Okanagan Specialty Fruits farms more than 1,250 acres of apple orchards in Washington, growing fruit that is sliced and sold to hotels and schools, and the company has been investing in forecasting technology to protect that output, Carter says. "You need to know more than just when your fruit is going to be ripe. How long do you have to pick it?" he says. "That's where these models are really helpful."

How does the camera-based system work?

Okanagan is testing cameras built by the Canadian firm Vivid Machines, mounted on top of tractors that drive through the orchard rows. As the tractor moves, the cameras collect imagery of the trees, and AI software identifies buds, flowers and fruit in that footage to estimate crop size and likely harvest dates.

"Right now, Vivid is telling us crop estimates and harvest dates," Carter says. He adds that the system is particularly good at spotting tiny flower buds that are difficult to see by eye.

But the forecasts are only as good as the data behind them, Carter says. "This isn't something where an AI can scrape the internet and figure out what's the average [yield] for Granny Smith," he says. "It's going to be bespoke to your farm."

Why don't experienced farmers just rely on instinct?

Asked whether experienced growers already know when their fruit is ready, Raymond Martin, co-founder and chief operating officer of the UK forecasting company FruitCast, says most do — but not evenly across an entire operation. "We do exactly what the farmers could do but we just do it on a scale that they can't," Martin says.

FruitCast builds its forecasts from footage of ripening fruit gathered by drones, a smartphone carried through a field, or a camera mounted on farm equipment. The company currently forecasts harvests for strawberries, raspberries, blackberries, blueberries and tomatoes.

What is the catch with AI forecasts?

The margin for error shrinks sharply depending on the crop. Apples offer some flexibility — Carter says the harvest window for Granny Smiths runs about three weeks — but soft fruit does not. "If a strawberry crop is on, you have to harvest it — otherwise your entire crop gets diseased very, very quickly," Martin says.

That urgency is part of why forecasting tools are being built crop by crop rather than as a single general model, according to both growers quoted by the BBC. A system tuned to a berry farm's microclimate and planting history cannot simply be copied onto an apple orchard, or vice versa.

How is the technology expanding?

FruitCast is already working across five fruit and vegetable categories and, according to Martin, plans to add forecasting for grapes next year. Okanagan's experience with Vivid Machines' tractor-mounted cameras points to a parallel path for orchard crops, where the near-term test is whether bespoke, farm-specific data can make bud and flower counts reliable enough to set labor schedules months in advance.

For now, the appeal for growers like Carter is narrower and more practical than any broad claim about artificial intelligence replacing a farmer's eye: a forecast that accounts for weather, crop condition and labor safety limits at once, in a season where a single hot morning can shut down picking before lunch.