Farm AI is finding an early home in office work

A new analysis of farmer survey responses places ordinary office work near the centre of early AI use. Terrain’s Marc Rosenbohm, writing on 26 October, found text tasks common and business analysis more prominent on larger farms. His analysis draws on Bushel’s 2026 survey, released earlier this year.[1]

Bushel’s April announcement covered more than 1,400 farmers in the United States and Canada. Among larger-farm respondents already using AI, 50% reported using it for business or financial analysis, compared with 25% for yield prediction or agronomy. Both percentages refer to AI users within that farm-size group.[2]

Those uses suggest a route into a working day that already involves correspondence, records and decisions about money. Consider an illustrative farmer asking for a clearer draft of a supplier email. The result is a document that can be inspected and edited immediately. A proposed change to field operations asks for a different kind of commitment.

A small age gap

Rosenbohm calculated an average age of 53.9 among AI users, against 55.3 for the whole sample. That 1.4-year gap compares users with all respondents, including some of the same people. It does not establish the difference between users and non-users, or show that age has no influence.[1]

The more useful commercial question is which job a product helps someone finish. A hypothetical budgeting assistant, for example, could produce an organised draft from figures supplied by the farmer. Its value would depend on whether the farmer can trace the inputs, correct the assumptions and use the result in an actual decision.

From an answer to an action

Purdue’s June Ag Economy Barometer asked 400 US farmers a broader question about AI or data-driven tools. Some 52% saw no meaningful benefit. Asked whether recommendations would be difficult to follow, 63% said sometimes and 22% said often. These are respondents’ views, rather than measured outcomes from a particular product.[3]

The Purdue and Bushel findings answer different questions in different samples, so their percentages cannot establish an adoption trend. Read alongside one another, they distinguish reported uses from willingness or ability to act on advice.[2][3]

That distinction gives the supplier-email example its practical force. A better draft can fit inside an existing task. A recommendation about running the farm must also fit the farmer’s resources and circumstances. A persuasive answer is only one part of that second job.[3]

Report an error or suggest a correction