AI · Drones · IoT for smallholder farms

Grow more. Lose less. Sell better.

AgriSmartOne helps Ghana's smallholder farmers decide what to grow, how to grow it and how to sell it, as one field journey: drones and satellites scout every zone, local AI reads the land, and the harvest is sold to the best bid, paid by mobile money.

Concept film Rendered scenes from our own 3D model library, not footage. The farm and the people are invented.
On the farm

One farm, one season, six jobs done for the farmer

What the drones, the sensors and the app do between planting and payment. Pick a scene to see it in the film.

Concept scenes Rendered from our own 3D model library, not footage. The farm, the people and the numbers are invented; yields and savings are design targets.

90%of Ghana's farms are smallholdings under 2 hectares
~20%of Ghana's GDP comes from agriculture
$12Bexport opportunity by 2030: cashew, shea, mango, pineapple, soybean
16demo farms running today on real drone surveys across Ghana
Market figures as cited in our solutions proposal; the demo farms are in the working app.
How it works

Find the problem, prove it, treat only that zone

Late disease detection, guesswork on inputs and middlemen setting the price cost smallholders most of their margin. AgriSmartOne closes the loop from the air to the market.

1

Drone scouting

Multispectral drones scan the crop for stress, disease and nutrient gaps, zone by zone; satellites watch between flights.

2

AI analytics

Imagery and sensor data become crop-health maps and recommendations, computed on the hub's own hardware.

3

Prescription map

Treatment zones with targeted irrigation and fertiliser plans, after a technician has confirmed the problem on the ground.

4

Precision treatment

Spray drones treat only the flagged zones, designed to cut chemical costs by 20–40%.

Never from imagery alone. A flagged zone is a question, not a verdict: a technician verifies it before any prescription, and advice is always for a zone, never the whole field.

The working app

Real screens, from the platform running today

These are screenshots of AgriSmartOne with its fictional demo farmer, Ama, and her 16 demo farms. Drone imagery is real; hardware, seasons and partners are simulated and labelled as such in the app.

The field journey

Twelve stages, from "which crop?" to "sold"

Each field has one home: a journey that knows what's done, what's next and when. Stages backed by real data complete themselves; the rest have a plain checklist.

  • Choose the crop by what it earns in this zone
  • Survey, plan, prepare, plant, scout, treat, harvest, store, sell
  • Guidance in plain language at every stage
Field journey · DEMO-01 · Maize
The field journey screen for a maize demo field, with twelve stages and the current stage's checklist
The field journey on a phone
Real screenFictional farmer and demo field
Whole-field land analysis

Every track, building, tree and weed patch, found by AI

From a 3 cm drone survey, the land analysis outlines farm edges, roads, buildings, trees, weeds and crop across every tile of the field, with counts and areas, on the hub's own computer.

  • Show or hide each class, measured in m² and hectares
  • "Grow more on the same land": price clearing stones, thickets and tracks
  • A track plan that gives land back to crops without cutting anyone off
Field · Drone land analysis
The drone land analysis panel over real drone imagery, with tracks, buildings and trees outlined and a list of classes with counts and areas
Real screen · real surveyNorth East Region, 3 cm drone imagery
Farmer dashboard

What needs attention, what it's worth, who's buying

The farmer's home shows the weather, overdue tasks, flagged zones, wallet balance, profit projections and the buyers asking for her crops.

  • Weather-aware advice ("hold off spraying until it passes")
  • Money from an immutable ledger, never a number someone typed
  • Buyers' offers next to the local price
Dashboard · Ama Farmer
The farmer dashboard with weather, tasks needing attention, money and buyer offers
The farmer dashboard on a phone
Real screenFictional farmer and figures
Market and money

Sell to the best bid, borrow from neighbours

Buyers post what they want and what they'll pay; farmers accept the best offer. Loans are funded peer to peer with the hub's pool, and repaid automatically from the harvest sale.

  • Buyer offers and auctions with mobile-money settlement
  • Sell-ahead forward and option contracts
  • Crop failure handled without punishing the farmer
Buyers market
The buyers market with offers from four buyers and their prices compared to the local price
Real screenFictional buyers and prices
The hub

One hub runs drones, labs and loans for a district

Hub staff see every field on the map, approve drone flights, soil tests and loans, and know which farmers need help today.

  • Every dispatch approved by the hub, charged at approval
  • Crops and demand across the hub's farms
  • Satellite health for every field, every day
Kumasi Pilot Hub
The hub dashboard with a map of every field, requests waiting, loans and crops in demand
Real screenFictional hub data
Farm Tour

Sixteen farms, three lenses: investor, government, farmer

A read-only tour plays every demo farm's simulated season over its real drone survey, with the numbers each audience cares about.

  • Read-only logins for investors and government
  • Realistic season videos where they've been rendered
  • A kiosk loop for demos and events
Farm Tour · Investor lens
The Farm Tour with a satellite map of Ghana and the demo farms pinned across the country
Real screenSeasons simulated
Proof that travels with the harvest

A field's whole story, that anyone can check

Lenders, buyers and inspectors ask the same questions: what has this field grown, how is it doing, and can I believe the records? AgriSmartOne answers each one from satellite data, the field's own history and records nobody can quietly change.

Field history

Ten years of every field, from satellite

Every season the Sentinel-2 satellite saw on the field since 2017: when it greened up, peaked and was harvested, how strong it was against the field's usual, what was probably grown and roughly what it was worth.

  • Radar fills in the cloudy months that cameras can't see
  • An alert when the field falls well below its own usual for the time of year
  • Crop guesses stay labelled as guesses until the farmer or hub confirms them
Field · History · DEMO-07 · Groundnut
Ten years of satellite greenness for a groundnut field in the North East Region, one peak per season from 2017 to 2026, with the seasons seen, the strongest season and the trend
Real screen · real satellite dataCopernicus Sentinel-2 over a demo field in the North East Region
Field Health Score

One number a lender can read, and open up

A score out of 100 built from parts anyone can see: greenness against the field's own past years, evenness from the drone, problems confirmed on the ground, and how well the crop suits the site.

  • Missing parts are left out, never scored low
  • The grade says what it rests on: site only, satellite, satellite + drone, or confirmed on the ground
  • Kept week by week, so a lender sees the trend, not one day
Field · Health · DEMO-07
The Field Health Score breakdown for a demo field: 87 out of 100, with greenness, evenness, confirmed problems and site fit, and the real satellite score beside it
Real screenDemo season simulatedThe real satellite score sits beside it
Digital Field Passport

Records no one can quietly change, us included

Everything done on the field is written down as it happens. Each record is chained to the one before it and signed, and once a day the day's fingerprint is sealed into Bitcoin through public timestamp services. A buyer, lender or inspector scans the QR code, and the check page tests every proof in their own browser.

  • A correction is a new record; old ones are never edited
  • The public page shows region, crop, area and the proofs, never names, prices or buyers
  • In the prototype the platform's server holds each person's signing key; in the pilot it moves to their phone
Field · Passport · DEMO-07
The Digital Field Passport for a demo field: a QR code to share it, and a verified seal saying all 22 records link up, match their hashes, are validly signed and are sealed in Bitcoin
The public passport check page on a phone, as a buyer sees it after scanning the QR code: the region, crop and area, and a verified seal
Real screensThe phone shows what a buyer sees from the QR code
Harvest lots and export

Four gates before a lot ships

The harvest is packed into lots that carry the passport with them. A lot is ready only when all four gates pass for the market it is going to. No border accepts a private record as a certificate, so the passport holds the evidence, pre-fills the official papers and links each one to the issuer's own check.

Gate 1

Field

The passport is intact and the field meets the market's rules, including a plot map for the EU's deforestation check.

Gate 2

Grade

The lot is graded at the packhouse against the market's standard, and the result is recorded in the passport.

Gate 3

Lab

Samples go to an accredited lab for residues and toxins, and the result is recorded against the lot.

Gate 4

Inspector

The official phytosanitary inspection, recorded with its certificate. Then the export dossier and the pack's QR code are ready.

Films

A whole maize season, from orbit to payment

A satellite scan flags a hot zone, a drone maps one farm, and the film follows its season: terrain, irrigation, zone-by-zone scouting, a zone-only spray, harvest, hub storage, the auction and the mobile-money payment.

The full seasonSimulated season · real drone imagery DEMO-01 · Maize · North East Region
Finding a disease, zone by zoneConcept film Close inspection, confirmation, a spray for one zone
The whole season on one phoneMark the plot, pick a crop by what it earns, call the drone, confirm a weak zone, sell to the best bid. Concept animation
Finding cocoa swollen shoot before it spreadsSimulated scene · real drone imagery DEMO-13 · Cocoa · Western North · 2:15, with voice-over. Hyperspectral detection is a trial with CRIG, not yet proven.
The same story in 60 secondsSimulated The survey, the AI's first look, the check on the ground and the replanting

Watch every video

Films, clips of the real app and animations, with playlists for farmers, hubs, government, buyers and lenders.

Local-first AI

Farm data stays on the farm's own hub

The AI that reads drone imagery and labels the land runs on computers in the hub, not in a foreign cloud. If the internet drops, the hub keeps working.

  • On-site inferenceSegmentation and labelling models run on the hub's hardware.
  • Never blocked by connectivityField work and analysis carry on offline and sync later.
  • Disclosed, alwaysAnything simulated is labelled; any outside service is listed in a register.
Business model and impact

Hubs that pay for themselves, farmers who earn more

How it earns

  • Drone services: survey and spray, per hectare
  • Software subscriptions for farmers, hubs and buyers
  • Soil and IoT testing
  • Export aggregation and a small platform fee

Investor value

Hub hardware is planned to pay back in about 22 months, with a national roll-out district by district. A planning estimate, not a forecast.

Impact

Empowering farmers, raising yields on the same land, and strengthening Ghana's agricultural economy, with the data kept in Ghana.

Bring AgriSmartOne to your district

We're setting up the first pilot hubs with farmer groups, ministries, lenders and buyers. Tell us where you farm, lend or buy, and we'll show you the platform on your own land.