Disrupting Agronomy? Mental Models for Why the Future of Agronomy is Ai-Augmented not Automated
Disrupting Agronomy? Mental Models for Why the Future of Agronomy is Ai-Augmented not Automated
In this Daily Scoop podcast, Bushel co-founder and President Ryan Raguse talks about AI in agriculture.
Three things stood out to me while listening to the podcast:
The Bushel view and what it might mean for their product feature development
What goes into a crop protection or fertilizer recommendation?
What will need to be true for LLMs begin to replace agronomists?
Bushel Product Development
Ryan suggests that AI will replace agronomist recommendations of fertilizer and crop protection recommendations in the next three years:
“I think a large amount of simple things like fertilizer and chemical recommendations will be automated by farmers that keep good data”
A few minutes later Ryan states that “human in the loop” will remain across most scenarios, so I do not think his statement was intended to mean absolute automation, but it does seem suggestive of future product features we will see in Bushel’s software suite in the next three years— agronomic LLM capabilities.
I think it’s worth unpacking replacement of agronomists vs. augmentation of agronomists though.
What goes into a crop protection or fertilizer recommendation and the workflow of fulfilling it?
A recommendation is not just one thing, it is a series of steps taken by an individual to assess, search, sell and deliver a better result to a farmer. Farmers in many instances do not do this because they have to prioritize time elsewhere, or lack the specialized knowledge required (eg: technical training, up to date product awareness etc) to inform the best decision.
I’d break out a recommendations into a three step process, with the intensity of each depending on the complexity of the problem and need:
Information gathering about the underlying needs — What is the problem that needs to be solved? What are the issues that need to be addressed? What are the goals of the farmer This might include scouting a field, soil testing, asking questions or looking at sensor data for example.
Searching for the right product, practice, rate or action to manage field or farm issues, opportunities or reach goals— Given the findings from the first step, what tools would be best for a farmer to use to attain the right agronomic, environmental and economic outcomes without hindering future optionality.