[AI Revolution](137) Croft: "Stop Farming by Instinct... Food Revolution with AI"

"Breaking Down High Agricultural Barriers"
Development of Smart Farm Control AI
Crop Growth Monitoring with Vision AI
Greenhouse Control and Cultivation Consulting Included

CEO Ryu Hee-kyung of Croft (right) and CEO (CTO) Lee Woo-ram posed together for a photo before an interview with Asia Economy at the branch office in Seocho-gu, Seoul. Photo by Heo Young-han

"In most industries, gaining about 10 years of experience qualifies one as an expert, but agriculture is an exception. At best, it's just having farmed about 10 seasons. Even if a strawberry harvest is abundant one year, farmers often don't know why."

Ryu Hee-kyung, co-CEO of Croft, which develops AI for smart farm control, spoke about the farming challenges she personally experienced. After graduating from Seoul National University's College of Agriculture and Life Sciences, she worked on smart farm projects at international organizations. To avoid staying only in theory, she returned to farming and took up lettuce cultivation. However, if the sunlight was even slightly insufficient or excessive, the lettuce seedlings would be damaged and often had to be discarded entirely. Ryu said, "Even smart farms that can control greenhouse environments are operated with consulting from expert growers. To move away from agriculture relying on human intuition, we determined that AI is necessary."

To integrate AI, Ryu partnered with Lee Woo-ram, co-CEO who majored in Electrical and Computer Engineering at the same university. The two united their efforts and participated in the 3rd Agricultural AI Competition held in the Netherlands in 2022. The competition was about using AI in lettuce cultivation to see who could generate the highest sales. The two founders ranked first in AI technology evaluation and third in sales. While their technology was recognized, they also identified areas for improvement. Lee explained, "Controlling only environmental variables like temperature and humidity has its limits. We concluded that cultivation strategies must be derived based on the crop's growth status."

Croft first developed a monitoring system for lettuce and tomatoes, which have high market demand. Vision AI observes the lettuce and collects data in 3D form. Based on this, it predicts weight, leaf area, growth rate, and yield with over 90% accuracy. They also created algorithms to check if crops are growing well and to control environmental variables if problems arise.

Next year, they plan to commercialize the technology. They are currently conducting proof of concept (PoC) with five strawberry farms and are preparing to open a demonstration complex in Gimje, Jeonbuk this spring. They plan to create a kind of pop-up store applying Croft's technology targeting startup farmers entering agriculture. This involves providing a subscription-based service (SaaS) with mobile robots equipped with vision AI that roam farms collecting data. Lee said, "As more crop data accumulates through commercialization, we can develop an AI system as skilled as expert growers. Ultimately, our goal is to develop an operating system (OS) where AI can observe crops, make judgments, and control the environment."

They are also targeting overseas markets. The Netherlands, a strong agricultural nation, has over 90% smart greenhouses, offering many opportunities to apply AI technology. With the rapid shift to facility cultivation due to global warming, they believe they can expand the market worldwide. Ryu expressed confidence, saying, "If we propose cost-effective cultivation strategies with AI, we can provide farmers with increased yields and consumers with price stabilization benefits."