KrishiSaathi
AI-Powered Multi-Agent Agricultural Intelligence
About the Idea
Brief About the Idea — Why AI is Required
Bridging the agricultural knowledge gap for 120M+ Indian farmers through multi-agent AI and voice-first multilingual design
🚨 The Problem
- 120M+ farmers lack timely access to expert agricultural guidance
- Information on diseases, prices, schemes is fragmented across platforms
- Language barrier — most solutions are English-only; farmers speak 22+ languages
- Low digital literacy — complex apps are unusable for rural farmers
- No personalization — generic advice ignores local soil, weather, crop context
- Delayed response — crop diseases spread while waiting for expert visits
💡 Our Solution: KrishiSaathi
- Multi-Agent AI — 5 specialist agents orchestrated by a Supervisor Agent
- Voice-first design — speak in any Indian language, hear back in the same
- RAG-powered knowledge — 50K+ verified agricultural documents for accurate answers
- Persistent memory — remembers farmer context for personalized advice
- Real-time data — live weather, market prices, and scheme updates
- AWS-native architecture — Bedrock, Polly, Transcribe, RDS, EC2
🧠 Why AI is Required
Traditional approaches cannot solve this at scale
AI enables real-time, personalized, multilingual agricultural guidance at a cost of ₹0.50/query vs. ₹500+ for an in-person expert visit. Multi-agent architecture provides specialized domain expertise that no single model can match. RAG ensures factual accuracy while LLMs provide natural conversation.
AI Disease Diagnosis
Instant crop disease identification from text/images. Treatment plans with costs and recovery timelines — no waiting for expert visits.
Bedrock Vision
Voice AI in 12 Languages
Amazon Polly neural TTS + Transcribe STT enables illiterate farmers to interact naturally in their mother tongue.
Polly + Transcribe
Contextual Memory
Mem0-inspired memory engine learns farmer's crops, location, soil type — making every interaction more personalized over time.
10 Categories
RAG Knowledge Base
50K+ verified agricultural documents across 8 specialized collections ensure factual, trustworthy recommendations.
ChromaDB · 8 Collections