← All food signals
Infrastructure

The Farm Memory

Deere and FBN are building farm AI assistants that draw on a farm’s own records. Both companies have also stated how the data is controlled.

What is it?

Farm AI assistants are being built to draw on an operation’s own records. Deere’s JD reads planting, yield-monitor and sprayer data; FBN and Google are assembling a context engine across agronomy, finances and equipment. Whether a stated data pledge settles who holds the memory stays open.

Why now?

A farm’s records are the ground these assistants answer from, so the question of who holds them arrives with the product. Deere issued a farmer data commitment alongside JD, and FBN’s co-founder says Google gets no access. The data statement now ships with the assistant.

Who’s building it?

An equipment maker and an input marketplace, working with Google’s fund, each build around farm data.

DeereFBNGoogle AI Futures FundOperations CenterEarly test users
The full read
Why now, who is building it, where the market goes, and what stays unsettled.
Get weekly →
01 · The market
Who’s in it
Builders — Deere and FBN, each building around farm data.
Buyers — farm operators in Deere’s app, and early test users of FBN’s features.
Gatekeepers — the data holders themselves, each stating its own commitments.
What’s at stake
Who holds a farm’s records once an assistant reads them: growers control, says Deere; FBN says it stays custodian.
An agent that assists with a purchasing decision after analysis.
Whether farm data stays out of AI training.
What’s unsettled
Which model sits under JD, since Deere declined to name it.
Whether a release limited to selected US operators widens.
How FBN’s early test features reach wider groups of farmers this autumn.
02 · How the demand matureda progression, not a timeline
Generic
Most tools offer broad advice, little tied to the farm, AgFunderNews notes.
Own records
JD reads planting, yield and weed-map data to answer for one field.
Purchase
FBN’s co-founder says an agent can assist with a product selection and purchasing decision.
Data statement
Both companies say how data is controlled; McKinsey ties adoption to farm-specific tools.
03 · Where the market goesthe forks, not a prediction
→Records or generic: assistants may keep drawing on each farm’s own data, or fall back on broad advice where records are thin.
→Deere’s staged release may widen beyond selected US operators, or stay with early users.
→Agents may assist with purchasing decisions after analysis, or stop at advice and leave buying to the farmer.
→Data commitments may become the standard farmers expect, or stay each company’s own statement.
The unsettled part is trust in custody: whether a company’s own statement about holding the records is enough for farmers to let an assistant read them.
04 · The Digital Sovereignty deskadjacent markets, moving now
Custody Over Cloud
Assistants sold on who owns context.
Town says customers own the context its AI assistants build
The Next Web
The Sovereign Model
Nations keep language, values and data at home.
Mistral releases an open-weight model with an EU data-jurisdiction option
AFP via Kuwait Times
The Modelled Harvest
Official crop figures move toward satellite models.
USDA plans an annual review comparing crop forecasts with final numbers
RFD-TV
05 · The Wire
01
Deere’s JD answers from a farm’s own records.
DTN ↗
02
Deere issued a farmer data commitment with JD.
Transport Topics (Bloomberg) ↗
03
FBN and Google build a context engine for farm AI.
AgFunderNews ↗
04
McKinsey: adoption depends on farm-specific tools.
RFD-TV ↗
The panel on this signal
They agreed on
–Farm AI assistants are being built on each operation’s own records, led by an equipment maker and an input marketplace.
–Each company has stated how the data is controlled.
–Both launches are staged, Deere’s to selected US operators and FBN’s to early test users.
The security researcher disagreed on
–A stated commitment is not an access control. Custody and control are different things; only enforced limits settle either.
Know more about the method →
✦How this signal was built
01 · Panel
A Delphi panel of 4 AI analysts, each reasoning independently.
→
02 · Debate
They challenge and revise each other’s reads over two rounds.
→
03 · Evidence
Every claim is tied to a cited, linked source on The Wire.
→
04 · Synthesis
A moderator writes it into one read, disagreement and all.