Work / Android apps / Pilot

Crop advisory app for farmers in regional languages

An Android app that lets farmers photograph a crop problem or ask a question by voice in Gujarati or Hindi and get practical guidance.

Problem

Farmers relied on visits from field officers for crop advice, and most apps were in English and assumed fast internet and new phones.

Objective

Give farmers quick, understandable advice on low-cost Android phones, in their own language, with an expert available when the app is unsure.

Solution

  1. 01

    Farmers photograph an affected leaf; an on-device model suggests likely issues, and a server-side vision model confirms when online.

  2. 02

    Questions can be asked by voice and answers are read aloud, in Gujarati or Hindi.

  3. 03

    Answers come from an approved advisory knowledge base, not the model's general knowledge.

  4. 04

    Low-confidence cases are sent to an agronomist, who replies in the app.

Architecture

01
Android app
02
On-device TFLite model
03
FastAPI backend
04
Vision model + advisory knowledge base
05
Speech-to-text / text-to-speech
06
Agronomist review queue

AI workflow

01

Farmer takes a photo or asks a question by voice

02

Run a first check on the device

03

Confirm with the server model and the advisory knowledge base

04

Reply in the farmer's language, as text and audio

05

Send uncertain cases to an agronomist

Technology & role

KotlinJetpack ComposeTensorFlow LitePython / FastAPIGemini API
Our role

Product design for low-end devices, Android development, AI pipeline and regional-language voice.

Start here

Want something similar built?

Send a short brief. We'll reply personally with questions or a suggested next step — usually a technical discovery call.