Industries / Food, farming and forestry / Agriculture and farming / AgTech
Farm software and data
This is the software farmers and their advisers use to keep records, plan work and get advice, and the rules on who controls the data it collects.
5
Milestones achieved, from 1975 to 2022.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1975
Farm sums by telephone
The University of Nebraska launched AGNET, a computer network that farmers and county extension agents reached by typing on a portable terminal connected through an ordinary telephone line. Its more than 200 programs worked out feed mixes, irrigation rates, marketing plans, loan costs and taxes for a farmer's own figures; by 1980 it had more than 1,100 registered users in 30 US states and two Canadian provinces, run jointly by six states, with user fees of about $10 an hour expected. The Christian Science Monitor gives 1975 as the launch year; a 1977 Daily Nebraskan report says the system came into use in 1974.
Credit
James Kendrick and Thomas Thompson, University of Nebraska
2012
Text-message prices, small gains
Reuters Market Light sold farmers in India market prices and weather forecasts by text message. A randomized experiment in 100 villages of Maharashtra found that farmers who got the service linked it to decisions they made, but found no statistically significant average effect on the prices they received, the value their crops added, losses to rainstorms or changes in crop varieties and practices; few farmers in the study districts were buying the service.
LessonFarmers using an information service is not evidence that it raises what they earn; that has to be measured.
Credit
Marcel Fafchamps and Bart Minten, World Bank Economic Review
2014
Farmers set data rules
In November 2014 the American Farm Bureau Federation, commodity groups and technology companies including Deere, CNH, AGCO, Dow, DuPont and Monsanto agreed the Privacy and Security Principles for Farm Data, which hold that farmers should own information from their farms and be able to move it between platforms. Because pledges alone did not check what companies did, the backers formed Ag Data Transparent, a nonprofit that reviews each applicant's farmer contracts against its yes-or-no answers to 11 questions, including whether a farmer can retrieve the original complete dataset, and renews the seal each year.
Credit
American Farm Bureau Federation; Ag Data Transparent
2019
Phone advice, measured
A review in Science pooled studies of farm advice sent to farmers' phones and found that it raised yields by 4% and the odds of adopting recommended inputs by 22% on average, with benefits likely more than ten times the cost of sending the messages. The authors added that messages written by agriculture ministries are often hard for farmers to use, and that services need feedback from farmers to be tested and improved.
Credit
Raissa Fabregas, Michael Kremer and Frank Schilbach
2022
A $300 million app splits
DuPont Pioneer bought Granular, a farm management software start-up that let farmers track costs and operations field by field, for $300 million in August 2017, by AgFunderNews's account. Its successor, Corteva, closed Granular Agronomy in August 2022 and on 10 November 2022 sold Granular Business, the farm financial software, to Traction Ag, saying "our ambition is not to sell our customers software". By October 2025 Corteva had turned what remained of Granular into a tool for its sales staff; farmers can still log in to see their data.
LessonA seed and chemical company can reshape the farm software it owns around its own sales, so farmers need a way to take their records elsewhere.
Credit
Corteva; Traction Ag
Watch for
A digital ID per farmer
India's AgriStack joins three registers kept by the states: farmers, georeferenced village maps and the crops sown on each plot. Each farmer gets a Farmer ID used for payments under PM-Kisan, crop insurance, minimum support price purchases, credit and disaster relief. India's agriculture ministry said more than 84.8 million (8.48 crore) Farmer IDs had been generated by 4 February 2026 and that the 2025 monsoon-season crop survey covered more than 285 million plots in 604 districts; Swarajya reported in October 2026 that the count had reached the target of 110 million (11 crore).
Achieved whenan independent audit or survey checks a sample of Farmer IDs against real farmers and land records and reports how many farmers received payments through them.
Claimed by
Ministry of Agriculture and Farmers Welfare, India, February 2026
AI advice by chat
Digital Green says its FarmerChat assistant, which answers farmers' questions by text and voice, reaches nearly 3 million farmers in six countries in more than ten languages, that about 70% of them act on the advice, and that the share of new users who ask a first question rose from 36.6% to 58.9% in India after a 2026 redesign.
Achieved whenan independent randomized evaluation reports yields or incomes for farmers offered the assistant against farmers who were not, or an outside group publishes a test of its answers' accuracy.
Claimed by
Digital Green, July 2026
Open
01
Exit test for farm records
A public test of 20 farm software platforms: enter the same invented farm (field boundaries, planting, fertilizer, spraying, harvest and costs) into each through free or trial accounts, export everything each platform allows, and check what survives and whether it loads into another platform.
The briefCredit · unclaimed02
Track farm data terms
A public archive that records every version of the data terms and privacy policies of the main farm software and machinery platforms, flags each change, and explains it against Ag Data Transparent's 11 questions.
The briefCredit · unclaimed
Open 01 · the brief
Exit test for farm records
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test of 20 farm software platforms: enter the same invented farm (field boundaries, planting, fertilizer, spraying, harvest and costs) into each through free or trial accounts, export everything each platform allows, and check what survives and whether it loads into another platform.
Why it is still open
Ag Data Transparent certifies companies on their contracts and their own yes-or-no answers, including whether a farmer can retrieve the original complete dataset; it does not run the export itself. The AEF announced in 2015 a conformance test for the ISOXML link between farm software and machines, which checks data exchange with machines, not whether a farmer's whole record moves from one platform to another. A 2023 University of Guelph study analyzed the text of 141 agricultural terms of use agreements, not the exports. A November 2026 search found buyer guides advising farmers to ask vendors, but no independent export test.
Built with what exists
Free and trial accounts; the open ISOXML libraries isoxml-py (Josephinum Research) and isoxml-js (dev4Agriculture) for the ISO 11783-10 machine data format; AgGateway's open ADAPT ISOXML plugin; GDAL to read shapefiles and GeoJSON; a browser agent such as Browser Use to repeat the export steps each month (see the Note "Deleting 80% to start over").
First step
Write a three-field invented farm as a spreadsheet and enter it into two platforms with free tiers.
Done looks like
A published table for at least 20 platforms showing, for each of eight record types, whether it exported, in what format and whether it loaded into a second platform, with the invented farm dataset and the method released.
Open 02 · the brief
Track farm data terms
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public archive that records every version of the data terms and privacy policies of the main farm software and machinery platforms, flags each change, and explains it against Ag Data Transparent's 11 questions.
Why it is still open
Open Terms Archive, an open-source project that records every version of digital services' terms, runs collections for generative AI, social media, Kenya, India and others, but in November 2026 none of its collections tracked a farm platform, and code searches of its archives for Deere, FieldView, Trimble, AGCO and CNH found none. The Guelph study collected its 141 agreements once.
Built with what exists
The Open Terms Archive engine; an AI model to summarize each change and map it to the 11 questions, such as Claude (see the Note "Seven more flagged", on a Claude-based document reviewer) or a local model run with ggml (see the Note "A first stable number for local AI").
First step
Write Open Terms Archive declarations for ten platforms' data terms and start the tracker.
Done looks like
50 farm platforms tracked for six months, with every change dated, shown word by word and summarized, each summary checked by a person, and the archive public.