136 builds1 posted in 2+ places23 Reddit · 107 GitHub · 4 dev.to · 2 AgOpenGPS forum
Models under the hood (where stated)
computer vision4
Gemini3
MobileNetV22
EfficientNet-B02
YOLOv112
Random Forest2
Mistral AI1
scikit-learn1
From the editor
Welcome to the first Field Radar. Farmers don't post "I built an app" the way coders do. They post a carrot that grew in a tube, a root cellar that took five months of block and grout, or a guidance app that runs on a phone and a $200 RTK receiver. This issue is all of that, 136 builds' worth, from the barn to the orbit above it.
The big story this month is satellites. Free imagery is now good enough that people are building field monitors on weekends. There's also more open-source precision ag than you'd guess, mostly on the AgOpenGPS forum, three separate goat apps (goat people, find each other), and a section on what the big companies shipped. Nobody, sadly, posted a combine. Maybe next month.
01
Lead story
Every field on Earth, watched from orbit, for free
The most upvoted ag build this month came from a coding subreddit, not a farming one. Someone vibe-coded an app where you tap any field on Earth and get a time-lapse movie of its NDVI, the crop-health signal agronomists pull from satellite images. It's free, and 149 people upvoted it.
It wasn't alone. A builder turned California's field-level crop records into a playful map of what's growing where. Another put live Sentinel-2 vegetation data behind a free farm monitor. A drought tool screens farms from orbit to decide which ones an inspector should visit first, and a fourth counts fruit and maps crop stress from drone and satellite images.
None of this is new science. What's new is that free imagery and free tools mean one person can now build in a weekend what used to take a company. Expect your agronomist's slide deck to change.
Drones & imageryAI coding assistant
Epic Fields
Tap any field on Earth and watch its crop health play back as a movie, straight from satellite. Free, vibe-coded, and the most upvoted ag build of the month.
Each of these is useful on its own. Hitch two together and you get something nobody on the forum has built yet. The builders have never met. Consider this the introduction.
agroguide-rtkOn its own: steers you down the field with a phone and an RTK receiver, but you set fields up in the cab.
+
AOG Field EditorOn its own: lets you draw and fix field boundaries and tracks in a browser, but doesn't drive anything.
→
TogetherFree phone-based guidance, with field boundaries you draw at your desk
Guidance in the cab, planning at the desk
AgroGuide RTK turns an Android phone and an RTK receiver into a guidance light bar with coverage painting. A separate builder made a browser editor for AgOpenGPS fields: boundaries, flags and tracks.
How it fits: If AgroGuide could read the editor's field files, you'd plan the field on a laptop and drive it from the phone.
smart-agri-lora-iotOn its own: puts cheap, SIM-free sensors across a field, but the data needs somewhere to go.
+
farmcoreOn its own: gives a farmer their own data hub, but it needs sensors to fill it.
→
TogetherCheap field sensors feeding a farm data hub the farmer owns, not a vendor
Sensors that report home
A Taiwanese project open-sourced low-cost LoRa field nodes and gateways that need no SIM card. FarmCore is an open, farmer-controlled data core with installable apps on top.
How it fits: The LoRa nodes as FarmCore's first sensor app: field readings landing in a database the farmer controls.
cropwat-skillsOn its own: works out crop water needs from weather data, then stops at a spreadsheet.
+
ha-spruce-irrigationOn its own: controls irrigation zones from Home Assistant, on whatever schedule you type in.
→
TogetherIrrigation amounts calculated the FAO way, then run automatically on your valves
Do the math, then open the valve
One builder automated the FAO CROPWAT method, which turns weather data into how much water a crop needs. Another wrote a Home Assistant integration for Spruce irrigation controllers: zones, flow and moisture.
How it fits: CROPWAT's numbers setting the run times in Home Assistant, so the schedule follows the science instead of a timer.
agro-weather-mcpOn its own: answers weather questions for any farm location, with no view of the crop.
→
TogetherTap a field and see its crop-health history next to the weather that caused it
Why did that field go yellow?
The NDVI movie app shows how a field's crop health changed over time. A separate agro-weather MCP server gives AI assistants compact weather tools built for farmers and agronomists.
How it fits: The weather server feeding the NDVI timeline, so a dip in crop health lines up with the dry spell or the hailstorm behind it.
Herd TrailOn its own: keeps herd records, but stops at the barn door.
TogetherHerd records and GPS tracking in one place
Records in the barn, tracking on the pasture
Herd Trail is a free record-keeping app a builder is shaping with farmers' feedback. A separate livestock platform tracks herds by GPS with geofences, records and auctions.
How it fits: Herd Trail's day-to-day records plus the platform's location data would give one view of each animal, in the barn and out on grass.
03
Parallel builders
Same idea, different builders
Three goat apps, four crop doctors, two farm databases, two irrigation plug-ins. Same problem, different continents, nobody comparing notes. Somebody start a group chat.
A goat farmer who is also an app developer built a herd manager for their own herd. Another builder is making Herd Trail, a free app for farmers. A third made a free calculator for goat kidding due dates.
If they talked: One good goat app beats three half-built ones. The calculator is a feature, the herd manager is the core, and Herd Trail has the feedback loop. Goat people, please find each other.
At least four builders made phone or web apps that diagnose crop disease from a photo: coffee, cacao and banana, cacao again in an Andes orchard simulation, a full crop-health platform in China, and a farm advisor in Tamil Nadu.
If they talked: A shared, open set of labeled disease photos would make every one of these more accurate. Right now each team is collecting its own.
Two builders independently landed on the same idea: farm data should live in a database the farmer controls, built on PostgreSQL and PostGIS. One built FarmCore, the other wrote up the design of an operator-owned farm dashboard.
If they talked: Comparing schemas now, before the two drift apart, would give farmers one open format to move data between tools.
136 builds, 88 audiences. Farmers, obviously. Then it gets specific: olive growers, goat keepers, Kashmir plots, people growing ginger on purpose. Tap one and the Explorer serves up everything made for them.
Four builds, all open source, all from the AgOpenGPS crowd: a phone that steers, a field editor in a browser, a run logger a farmer had Claude write, and ISOBUS running on a ten-dollar chip.
Precision & guidance
agroguide-rtk
A free Android app that turns a phone and an RTK receiver into a guidance light bar, with coverage painting and NTRIP corrections.
Cheap boards, LoRa radios and a lot of soil moisture. Plus a ₹7,600 kit that babysits onion sheds and one honest write-up of a Kenyan deployment and everything that broke.
Sensors & IoT
IoT water sensors in Kenya
IoT water sensors went onto a farm in Kenya. This is the honest write-up of everything that broke, and what finally worked.
The lead story has the satellites. Still here: a Nigerian state's greenness over the years in Earth Engine, and a map that scores Azerbaijan by the acre.
Drones & imagery
oyo-ndvi-crop-monitoring
Crop greenness and rainfall across one Nigerian state, year over year, in Earth Engine. A drought shows up before anyone reports it.
Goats had a big month: a herd manager, a kidding calculator, and a records app shaped by farmer feedback. Also milk records that work without signal, and an egg incubator that used to be a smoker.
Livestock & dairyAI assistant
DairyLog Online
The offline dairy record app grew up: there's now an online version for teams running more than one farm.
Hydroponics owned the month. Carrots in tubes, strawberries in year seven, better net cups off a 3D printer, and crop steering wired into Home Assistant.
Greenhouse & hydroponics
Better 3D-printed net cups
Redesigned 3D-printed net cups for hydroponics, and the builder asked growers what to fix next. Forty of them answered.
The paperwork side: a data hub the farmer owns, a season of calculators that work offline, and an app that tells a Kashmir farmer what to ask the ag office.
Farm business & data
farmcore
A farm data hub the farmer owns, with installable apps on top. The vendor never sees it.
Sixty-one builds and half of them are somebody's thesis. The four here actually do something: a crop doctor that works offline, a living list of crop models, a simulated farm for robots, and flood warnings by text.
AI & research
healthycropping
Point a phone at a coffee, cacao or banana leaf and get a diagnosis, with no internet needed. Built for the places that don't have any.
While the garage builders were welding, the majors were busy: a Deere parts warranty, a new corn head, an autonomous Kubota, and a hundred spot sprayers in onion fields. Via Precision Farming Dealer.
