When can I harvest after the first bloom? And when will that harvest peak? These are questions growers are currently struggling to answer accurately in the absence of precise harvest forecasts.
Before the advent of Google Maps, motorists faced similar uncertainties. Traveling from Amsterdam to Rotterdam? It took about 50 minutes, but only without traffic jams. Nowadays, navigation software is so advanced that the expected time of arrival is continuously adjusted while driving. Even someone planning the same trip a month in advance can get an accurate estimate of the travel time.
Source.ag is achieving something similar with its latest product, Source Cultivate, but for tomatoes. Co-founder Rien Kammen draws this comparison. The fast-growing AI software developer enables growers to accurately predict tomato harvests up to eight weeks in advance.
A new puzzle piece
"To be successful, a grower must be able to anticipate," says the CEO. Once the fruits start to color, harvest forecasting becomes feasible, but before that, it's largely guesswork. This makes it challenging to plan labor, logistics, and sales. "The more you know about the future, the better. Source Cultivate, in our view, is the missing piece of the puzzle. An accurate harvest forecast with one clickâthat's what we make possible with this new product."
Ernst van Bruggen, also a founder and the chief product officer, emphasizes that predicting a tomato harvest up to eight weeks in advance has been "almost impossible." "Until now, it was impossible to accurately calculate everything due to varying weather and greenhouse climate conditions, among other factors. For us, this is now possible because we build on the knowledge and systems we already have."
The fast-growing company first digitized the greenhouse from Amsterdam with Source Track and the Source Plant App. It then launched Source Irrigation Control this spring, an intelligent AI-based software that gives growers a co-pilot to autonomously control irrigation.
Rien and Ernst, founders of Source.ag
Simulation
Source Cultivate also uses smart AI software that processes a wealth of data already available from growers. The system operates on the basis of simulation. Ernst explains, "We simulate cultivation from flowering to harvest and update everything daily. For this, we use data from the greenhouse and the climate computer. With our Plant App for plant registration, growers record, for example, the number of fruits added to a plant each week. We need that data for accurate predictions."
Rien emphasizes that Source.ag opts to work with all the data already available, without requiring a separate camera system in the greenhouse. "We seek the most pragmatic and cost-efficient solution, using all available data. If future data includes advanced imaging technology, we will definitely integrate it."
More certainty every day
Currently, about 50 growers are already working with the AI harvest forecast on a trial basis. All are tomato growers, as Source.ag is introducing Source Cultivate for this crop first. When forecasting, the system also uses variety data and plant physiological knowledge accumulated over the years. However, the weather is not so easily predicted.
"For us too, weather is a challenge," acknowledges the Chief Product Officer. "However, the advantage of our approach is that if there is a heat wave, for example, the entire system makes an immediate correction automatically. If you anticipate something, you can factor it into the calculations. And if that fails, the system continues to calculate as soon as changes occur. Every day, the uncertainty about the harvest decreases, and the certainty increases."
During the test run, the software developers discovered that Dutch growers in their high-tech greenhouses can already predict their winter harvests quite well. Rien says, "Then they have a lot under their own control. The great strength of forecasting with Source Cultivate is that even for those ten hard-to-predict weeks in the year, it will now be possible to accurately predict the harvest."
Click here to enlarge this screenshot from Source Cultivate
90% accurate
The key question, of course, is how accurate Source Cultivate's prediction is. Ernst says, "The accuracy depends on the variety. The more data, the better. On average, for all the 40 or so tomato varieties we now have experience with, and there are more and more, we achieve a 90% accurate prediction of the harvest, 3 to 4 weeks ahead, if the grower follows the cultivation plan."
The software developers aim to make things easier for growers with their products. Rien adds, "With Source Cultivate, we help growers with their planning, and it also supports them in discussions with supermarkets about the price of their product."
A better harvest forecast is also very welcome further down the chain. That is why Ernst says it is important that Source Cultivate also solves the second half of the 'harvest forecast problem,' which is communication. "Through an API, Source Cultivate communicates directly with the ERP system of the sales organization that sells the tomatoes for the grower."
Besides being a forecasting tool, the software also serves as a feedback tool, according to the CPO. "At the cooperative level, for example, you can start looking at how to optimally align the harvests of different growers to avoid surpluses and shortages." Rien concludes, "Therefore, at the cooperative level, we believe it is truly a game-changer."
At GreenTech Amsterdam, Source.ag can be found at stand 01.545 on 11, 12 and 13 June. There will be the opportunity to get a demo of Source Cultivate.
For more information:
Source.ag
[email protected]
www.source.ag