Gona: a crop doctor that works where there's no signal

Description and use

Gona is an offline crop‑diagnosis app that uses a local speech‑to‑text model and a static knowledge base to provide farmers in remote areas with timely disease identification and treatment advice.

Problem

Farmers in remote West African fields lack reliable internet, making existing crop diagnosis apps unusable.

Outcome

Gona delivers accurate, offline crop disease diagnosis for nine crops and eleven conditions, with no internet required and zero inference costs.

Workflow

• 1. Farmer opens Gona app on phone. 2. Presses the single button and speaks symptoms. 3. Whistle (16.9 MB WebAssembly speech‑to‑text) transcribes speech locally. 4. Symptom matcher compares transcription to a JSON knowledge base of 11 conditions. 5. App reads out the matched diagnosis and treatment steps. 6. If no match, app informs the farmer to consult a local extension service.

Evidence

Every crop diagnosis app I've tried assumes you've got signal in the field.

Source

It runs in the browser tab as WebAssembly. After your first visit, a service worker caches the app, the knowledge base, and the model runtime.

Source

The only neural network in Gona is Whistle, a 16.9MB open-weight speech-to-text model from Cactus Compute (Apache 2.0).

Source

Whistle has one job: hear what you said. Everything after that is a lookup against 11 conditions I pulled from published plant pathology and extension sources (CABI, IITA, APS).

Source

I tested the matcher against 12 field-style descriptions, written the way a farmer would actually say it, not in textbook language.

Source