Source.ag improves tomato harvest forecasting AI

Source.ag announced updates to its AI-powered harvest forecast model for tomatoes, reporting gains in accuracy and reduced data entry requirements for greenhouse growers.

The next-generation model introduces changes in how it is trained and updated, using aggregated cultivation data to better account for climate variability and greenhouse-specific conditions. According to the company, the model improved mean forecast accuracy by 33% at a three-week horizon compared with the previous version.

The share of forecasts with more than 20% deviation declined by 25%, while severe forecast errors were reduced by 50%. These improvements may help growers better align production planning, labor and sales commitments.

The updated system also automates several processes that previously required manual data entry, reducing operational workload while improving forecast performance.

The rollout is underway, with initial greenhouse operations already using the model and broader deployment expected in the coming months for tomato growers on the platform.

“What makes this meaningful is how the model improves over time,” said rien kamman, CEO and co-founder of Source.ag. “The AI model learns from every cultivation running through Source.ag, so the model’s performance compounds as more growers use the platform.”

For greenhouse operations, more accurate forecasting tools can support crop planning, reduce waste and improve pricing strategies in a variable production environment.