Rows of butter lettuce grows in a controlled greenhouse
An arrow pointing leftHome

Seattle startup Koidra is tapping AI to automate greenhouses

  • Louise Matsakis

The company has raised $4.5 million in seed funding.

Global food prices have skyrocketed over the last few months, making it harder for many families to make ends meet. Koidra, a Seattle-based startup founded by former Microsoft research scientist Kenneth Tran, wants to help by using artificial intelligence to automate greenhouses, reducing waste, and improving crop yields over time. Earlier this year, Koidra announced it had raised $4.5 million in seed funding.

The startup is part of the rapidly growing agriculture technology market, which has attracted significant attention from investors recently. Venture capital firms poured $10.5 billion into agritech startups in 2021, a more than 58% increase from the year prior, according to data from the research firm PitchBook. Some of the new firms are based in the Seattle area.

Making agriculture more efficient is an urgent priority in many parts of the world, since the global population is expected to reach 9.7 billion by the middle of the century. Feeding all those people will be a tremendous challenge, especially as climate change makes farming more difficult and unpredictable. The COVID-19 pandemic also demonstrated that current global food supply chains are vulnerable to disruption, helping spur investment in new agriculture technologies.

One way to shield crops from hotter temperatures and depleted soil is to move them into greenhouses, where conditions can be carefully regulated with software. Many greenhouses specialize in hydroponic crops, which are grown in a mineral-rich liquid solution instead of soil. There are already more than 2,300 hydroponic farms in the United States, according to The New York Times.

In an interview with GeekWire, Tran, Koidra’s founder, said he first became interested in indoor agriculture when he noticed a vertical farm in Microsoft’s offices. “I got hooked into the problem of indoor farming. It’s a research friendly project,” he told the tech news site.

Tran later began competing in international greenhouse competitions as part of his work at Microsoft. In 2018, he and a team of researchers won the Autonomous Greenhouse Challenge for growing cucumbers, which they managed remotely using AI. The judges said the fruits were superior to those cultivated by expert human growers.

Tran left Microsoft to start Koidra in 2020. The startup went on to win the same greenhouse competition again this year, only this time for cultivating lettuce. The contest lasted four months, during which Tran and his team used machine learning to control conditions in the greenhouse, like the amount of ventilation, heating, and irrigation.

Tran specializes in a specific type of machine learning called reinforcement learning, which involves teaching an AI how to respond to complex events in real time. The same tech has been used to train self-driving cars and to teach artificial intelligence how to play video games. Koidra says it wants to use reinforcement learning to automate other types of facilities, such as factories and power plants.

Most of Koidra’s current clients are in markets like the United States and Canada, but the company is also expanding to Vietnam, where Tran’s family lives. He told the news site Tech in Asia that he wants to “democratize” Koidra’s technology and ensure its affordable enough to be used in developing countries that currently rely largely on small-scale farming.