Skip to main content
All Case Studies
AgriTech KhetSaathi (AgriTech, India) India 5 months

AI Farmer Advisory Platform for AgriTech Startup

KhetSaathi had grant funding to build a farmer advisory platform for Maharashtra and Karnataka, providing crop disease diagnosis, weather-based alerts, and mandi price information to smallholder farmers.

67% (from 3%)
Monthly active users
78% of interactions
Voice message usage
89% (field validated)
Crop disease diagnosis accuracy
8 seconds
Avg response time
52,000 in 4 months
Farmers onboarded
< 50KB (2G compatible)
Data usage per session

The Challenge

KhetSaathi had grant funding to build a farmer advisory platform for Maharashtra and Karnataka, providing crop disease diagnosis, weather-based alerts, and mandi price information to smallholder farmers. Their target users: 50,000 farmers in Tier 4 areas with 2G connectivity, low literacy, and no English proficiency. Their previous app, built by a different agency, had 3% monthly active usage after 6 months.

Our Solution

Canny rebuilt the platform as a WhatsApp-first chatbot (no app download required) with Marathi and Kannada voice message support, photo-based crop disease diagnosis using a custom CNN model, and integration with IMD weather data and AGMARKNET mandi price feeds. Monthly active usage reached 67% within 4 months.

Technical Architecture

WhatsApp Business API via Meta directly (high volume)

Node.js serverless backend on AWS Lambda

Custom CNN model for crop disease detection (TensorFlow, trained on 120,000 labelled images)

Bhashini API for Marathi/Kannada ASR (voice to text)

GPT-4o with RAG for crop advisory (30,000-entry knowledge base)

AGMARKNET API for real-time mandi prices, IMD API for weather

Technologies Used

Node.jsAWS LambdaWhatsApp Business APITensorFlowGPT-4oBhashini APIAGMARKNET APIPythonPostgreSQL

Project Timeline

1

User Research and Technical Architecture

3 weeks
  • Field interviews with 40 farmers in Marathwada and North Karnataka
  • Connectivity and device testing in target areas (2G/3G)
  • WhatsApp as primary channel decision (over app rebuild)
  • Offline-tolerant architecture design
  • Bhashini ASR language model evaluation
2

Crop Disease Model and WhatsApp Integration

8 weeks
  • TensorFlow CNN training on 120,000 labelled crop disease images
  • WhatsApp Business API integration for photo and voice message handling
  • Bhashini Marathi/Kannada ASR integration
  • GPT-4o advisory pipeline with crop-specific RAG knowledge base
  • Marathi/Kannada TTS for voice advisory responses
3

Data Feeds and Farmer Onboarding

6 weeks
  • AGMARKNET mandi price feed integration (18 mandis)
  • IMD weather API for district-level 7-day forecast
  • PM-KISAN beneficiary API for farmer verification
  • IVR onboarding flow for feature phone users
  • Field agent training in 200 villages

“Farmers send a photo of their crop with a voice message in Marathi and get an advisory in 8 seconds — in Marathi voice. No app, no English, no data plan needed. That's why they use it. Our previous app required Wi-Fi and English. Nobody used it.”

— Dr. Suresh Waghmare, CEO — KhetSaathi

Long-Term Impact & ROI

KhetSaathi's platform achieved NABARD endorsement and was selected as a pilot for PM-KISAN advisory integration. 78% of surveyed farmers reported reducing pesticide use based on the diagnosis recommendations. The platform attracted ₹12 crore Series A funding.

Want Results Like This?

Every case study started with a free discovery call. Let's explore what's possible for your business.

Book Free Consultation

More Case Studies

Marketing Technology

AI-Powered Marketing Automation Platform

GrowthMatic (B2B SaaS) · USA & Canada

47hrs Hours Saved Per Week−78% Campaign Launch Time
Government Services / Consumer Tech

AI Passport Photo Processing Platform

PhotoID Pro · UK & Europe

15,000+ Photos Processed Daily97.2% Government Acceptance Rate