Logo Lanfrica

Purvakapadia04/kisanmandi

Domaine:

agriculturesocioeconomic

Type de record:

software
Créateur:
Pur
Hôte:
AI-powered market intelligence platform helping Gujarat cotton and groundnut farmers compare mandi prices, find buyers, and make smarter selling decisions. # KisanMandi AI — Agentic AI Market Intelligence Platform > **Smarter Markets. Better Prices. Empowered Farmers.** An Agentic AI-powered platform built for **IBM TechXchange Hackathon — Challenge 13** that connects Gujarat's cotton and groundnut farmers directly with real-time market intelligence, verified buyers, and personalized AI selling recommendations. --- ## 🌾 Problem Statement Gujarat is one of India's largest producers of cotton and groundnut. Farmers face: - Price exploitation by middlemen - Lack of access to real-time mandi prices - Difficulty comparing prices across markets - Limited bargaining power - Inability to find genuine buyers - Uncertainty about when to sell vs store --- ## 💡 Solution — KisanMandi AI KisanMandi AI is a **multi-agent AI platform** that provides: 1. Real-time mandi price intelligence across 8+ Gujarat mandis 2. AI-powered 7-day price forecasting 3. Direct farmer-to-buyer marketplace with AI matching 4. Storage vs selling decision advisor 5. AI-assisted crop quality grading via image upload 6. Personalized income and profit calculator --- ## 🤖 Agentic AI Architecture The platform uses **IBM Granite LLM** as the reasoning core, orchestrating 5 specialized AI agents: ``` Farmer Query ↓ AI Orchestrator (orchestrator.js) ↓ ┌─────────────────────────────────────┐ │ Agent 1: MandiPriceAgent │ → Current prices, trends, forecasts │ Agent 2: BuyerMatchingAgent │ → Finds compatible verified buyers │ Agent 3: StorageAdvisorAgent │ → Sell Now vs Wait vs Store │ Agent 4: QualityGradingAgent │ → Image-based crop quality estimate │ Agent 5: IncomeAgent │ → Revenue, cost, profit calculations └─────────────────────────────────────┘ ↓ IBM Granite LLM Reasoning Layer ↓ Personalized Farmer Recommendation ``` --- ## 🏗️ IBM Cloud Architecture ``` Frontend (React/Vite) ↓ Backend REST API (Node.js/Express) ↓ Agent Orchestrator ↓ IBM Granite (WML - Watson Machine Learning) ↓ IBM Cloud Obj …