At last month’s World Agri-Tech summit in San Francisco, speakers framed the next phase of AI in agriculture as something new: systems that don’t just offer advice but begin acting on it.
Organizers themed this year’s summit the “agentic age,” a term gaining traction in tech circles.
WHY IT MATTERS: AI tools are improving quickly, but understanding where they fit in decision-making helps farmers use them without giving up control.
In simple terms, “agentic” refers to systems that move beyond analyzing information or making recommendations and begin taking on tasks and acting more like digital “agents.”
Ranveer Chandra, chief technology officer of agri-food at Microsoft, hosted the opening session at the World Agri-Tech summit and put the question cleanly:
Read Also
The wow factor: how dairy farming technology has changed in one lifetime
Lee Hart looks back at how dairy farming technology has scaled — from 20 cows in his father’s barn to 100,000 head.
“What does it mean for farmers to have a digital assistant or be managing a set of agents that are doing work for them?”
While the question is cleanly put, the answer is murkier.
It’s easy to see how AI is working in corporate settings. During the same session, Reza Rasourpour, a vice-president with Corteva Agriscience, pointed to how artificial intelligence is massively speeding up product development.
Those advances are making their way to the farm.
Rasourpour also spoke about a new Corteva fungicide timing system that uses artificial intelligence and field-level data to predict disease risk and identify the optimal spray window.
Reza Rasourpour of Corteva Agriscience speaks at the World Agri-Tech summit in San Francisco, where he highlighted an AI-driven fungicide timing system showing five to 10 bushel-per-acre yield gains. Photo: World Agri-Tech Summit
“We’ve seen upwards of five to 10 bushels per acre improvement when that fungicide application happens at the right time,” Rasourpour said.
But beyond examples such as fungicide timing, the panel discussion stopped short of exploring what a fully “agentic” system would look like in day-to-day farm management.
From concept to field
The pitch for AI in agriculture at events like World Agri-Tech is bold. The reality on Prairie farms is more cautious.
That’s where the question shifts from the conference stage to the field.
On Prairie farms, where decisions are shaped by weather, field variability and farmer risk tolerance, agronomists say there is still a clear line between a recommendation and a final call.
Even when farmers work closely with trusted advisers, they generally keep control over decisions that affect input costs, crop risk and yield potential. That is unlikely to change quickly just because the recommendation is coming from software instead of a person.
Rob Warkentin, a Saskatchewan-based private agronomist, says some AI applications already look realistic, especially in areas such as disease forecasting.
“AI is a great fit when it comes to things like disease,” he said. “But maybe not so much on fertility recommendations.”
That distinction is important.
Disease risk, insect movement and weather-driven threats are all areas where more data, better pattern recognition and faster analysis could improve timing. In those cases, AI may help narrow down a decision faster than an agronomist or farmer working alone.
However, that is not the same as handing over control.
“Even with my recommendations, they still want to have the final say,” Warkentin said. “And they trust me.”
He’s less certain computer models will ever gain that same level of trust with farmers.
That same skepticism came through in Manitoba.
The limits of data
Brunel Sabourin, co-owner of Antara Agronomy in St. Jean Baptiste, Man., says AI will be disruptive and already has obvious uses in helping agronomists and farmers process more information faster. However, he also sees hard limits in a business where no two seasons are the same.
“The biggest overarching challenge that I see is being able to capture all of the variability in a field and being able to make proper decisions with that,” he said.
For Sabourin, that variability is the reason agronomy still resists full automation. Soil, weather, moisture, field history and management all interact. A small change early in the season can ripple through everything that follows.
“Agronomy is an art as much as it is a science,” he said.
That does not mean AI has no role. In fact, both agronomists see it becoming more useful, not less.
Sabourin says his own business is already using AI heavily for lower-level tasks such as reports and analysis, and increasingly to dig through large agronomic datasets more quickly.
He described using it to sort through benchmarking data and test relationships between variables that would have taken far longer to examine manually. That, in itself, is valuable.
The local advantage
Attendees at the World Agri-Tech summit in San Francisco hear about the “agentic age” of AI — a vision Prairie agronomists say still has to clear the bar of real-field variability. Photo: World Agri-Tech Summit
But even there, Sabourin cautions against assuming that more data automatically leads to clean, scalable answers.
From the beginning, he said, Antara has approached agronomy benchmarking with what he calls “a small data approach,” comparing farms within a small local radius to avoid overstating what broad datasets can explain.
That local reality, from soil type to hyper-local weather patterns measured in-field, may be one of the clearest limits on how far AI-driven decisions can be generalized in Prairie agriculture.
“The way that I see it is the AI is just a tool to strengthen my recommendations,” Warkentin said.
“I don’t know that any of these things are going to remove a person from the field, but they will make us better at going to the right parts of the field to determine what’s happening in different areas.”
