Agriculture robots gather more than just crops — they harvest data

Often overlooked in the ongoing conversation around AI, agriculture has been quietly experiencing its own technological revolution. Processes that took months have been shaved to a week, swaths of data are now available to help optimize crop rotations and profit. The magnitude of efficiencies is now drawing the attention of new investors and entrepreneurs. 

We spoke with Walt Duflock, VP of Innovation at Western Growers, a leading non-profit representing the interests of family farmers growing fresh produce in Arizona, California, Colorado, and New Mexico, to discuss the tipping point agriculture has reached for unthinkable efficiencies and new business potential. 

Agricultural robotics and AI: According to Duflock, agricultural robots have benefited enormously from AI advancements. "Agricultural robots you see out there are benefiting from AI in all forms and sizes," Duflock explains. "If you look at weeding robots, they went from not being able to weed any crops five or six years ago to now, in Carbon Robotics' case, weeding 100 different crops. This happened because AI enabled the learning process. The acceleration of capabilities has created a snowball effect that's rapidly gaining momentum."

From months to a week: The speed of this technological advancement is remarkable. Duflock recounts a recent conversation with the leadership of Carbon Robotics, who told him they can now add new crops to their weeding library in just one week—a process that previously took months.

"They can take a crop that they need to be weeding on Monday, do the imagery, do the passes through the field, and by Friday that crop can be in their library of crops to weed," Duflock says. "If you knew what that took seven or eight years ago, it was like months."

This rapid learning capability extends across the industry, from mechanical weeders like those from Stout and Farm Wise to spraying systems from companies like Ecorobotix and Guss. Once the AI can identify one crop, adding more becomes increasingly efficient.

Internal Value: Beyond robotics, Duflock highlights how companies are using AI internally to create value. He cites examples like Syngenta's Crop Wise system, which outperforms general AI engines like ChatGPT in specialized agricultural applications.

"If you have a bunch of content around a specific use case, like in this case, inputs, but it could be genetics or something else, you should be able to outperform a general AI engine," he explains.

Duflock believes the first areas to capture value from AI will be these internal use cases, where tech can be first implemented and tested for value versus being marketed directly as a customer-facing product. 

Agricultural Data Goldmine: Perhaps the most exciting potential, according to Duflock, lies in the massive amounts of field data being collected by agricultural robots. "All of these robots making all these passes through the fields gives us an absolute treasure trove of agronomic and plant health and field data," he says.

The next opportunity, Duflock suggests, will be for entrepreneurs who can aggregate and analyze this data to provide valuable insights to growers. He envisions a system that could recommend optimal crop rotations and planting decisions based on market conditions and historical performance.

"Imagine if you knew across a guy's 5,000-acre portfolio where the best places to grow broccoli are, where the best places to grow bell peppers are, and you knew when the market conditions suggested you should plant bell peppers in the next four days—that would be pretty valuable," Duflock explains.

New Venture Capital Opportunities: This data-driven approach also creates different funding dynamics for agTech startups. While some companies require significant capital investment—"the traditional $50-100 million to go raise and get out to market," as Duflock puts it—software and analytics startups can operate with a leaner model.

"I think the need for venture capital on the startup and analytics side is going to be way less," Duflock predicts. "I think we'll probably see a lot of $25-50 million exits, on that side of the business. When you don't take $50 million in venture capital, a $25 million exit can be more than fine."

Young entrepreneurs: Duflock believes there's a significant opportunity for young entrepreneurs from agricultural programs at universities like Cal Poly, Fresno State, Davis, and Purdue to develop these analytics platforms, creating business models where growers who contribute data receive benefits like revenue sharing or preferential pricing.

As robotics companies continue to scale, this opportunity becomes increasingly accessible. "Until the robots were out there at scale, you really couldn't do this as cheaply as you can now," Duflock concludes, pointing to a future where agricultural data becomes a valuable business in its own right.