Prototypes and pilots
Everything we have built so far, and where it goes next. Four prototypes prove that each part works. The first field pilot will test whether farmers act on the advice.
How we test our work
Ground truth first
No nutrient reading shapes advice until it has been checked against lab tests for that soil.
Built for the worst week
Radio links are planned for maize at tasselling in monsoon humidity, when signals are at their weakest.
Fits a real phone
Every instruction is tested against the 160-character, 150-second limits of a live USSD session.
Honest about unknowns
We say what we have not proven yet, such as whether farmers trust a voice message over community elders.
How we measure results
- 01
Do farmers hear it?
Advice arrives as a voice call. We count how many calls are answered and listened to for at least ten seconds, and call each farmer at the hour they usually pick up.
- 02
Do they act on it?
Every call ends with "Press 1 if you have done this." We count it for each message, and test new wordings of the same advice on half the farms to see which one more farmers act on.
- 03
Does the harvest change?
At the end of the season each farmer tells us how many bags they harvested. We compare farms that followed most of the advice with farms that did not, crop by crop.
- 04
Are farmers paid?
For produce sold on the platform we count offers accepted, payments that reach the farmer, and disputes, buyer by buyer and transporter by transporter.
These are comparisons, not proof. Farms that follow advice may differ in other ways, so a small gap, or one from a few farms, is treated as no gap. The team sees each number as it comes in, from the first pilot season on.