Kissan AI

setting and notification page
main chat page
cutomize location wheather dashboard
Profile page
Help Page

Kissan AI — Intelligent Agricultural Assistant for Pakistani Farmers

Track: MemoryAgent

The Problem 8+ million smallholder farmers in Pakistan face:

No personalized agricultural advice
Language barrier (most AI tools only speak English)
No memory across sessions — "Who are you? What do you grow?"
Reactive instead of proactive — find out about problems too late
No local market data for best sell timing

The Solution Kissan AI is a multilingual, memory-powered agricultural assistant that:

✅ Remembers every farmer's profile, crops, and history (MemoryAgent)
✅ Speaks Urdu, Hinglish, and English automatically
✅ Proactively alerts about weather, pests, and market prices
✅ Uses multi-agent collaboration (Crop, Weather, Market, Pest agents)
✅ Trained on Pakistani agriculture knowledge (PARC, Punjab Agri Dept)

Key Features

Multi-Layer Memory System:
Profile Memory: Name, location, land size, soil type
Session Memory: Current conversation context
Long-term Memory: Historical farm data
Episodic Memory: Past interactions and outcomes
Semantic Memory: Learned best practices
Multilingual Intelligence
Auto-detects English, Urdu (اردو), Hinglish
Responds in the same language
Voice input support
Multi-Agent Collaboration;
Crop Agent: Disease diagnosis, cultivation advice
Weather Agent: Forecasts, spray timing
Market Agent: Mandi prices, best sell timing
Pest Agent: Pest outbreak warnings
Proactive Alerts
Weather: "Barish aa rahi hai — aaj hi spray kar lo"
Pest: "Sundhi attack risk high this week"
Market: "Cotton rate up in Multan — sell now"

Tech Stack

Frontend: HTML5, CSS3, JavaScript (mobile-first)
Backend: FastAPI (Python, async)
AI/LLM: Qwen Cloud API 3.5/3.6 (Alibaba Cloud)
Database: SQLite (farmer profiles, conversations, memories)
Vector DB: FAISS (RAG for agriculture PDFs)
Weather: OpenWeatherMap API
Deployment: Render

Impact

Target: 8+ million smallholder farmers in Pakistan
Scalable to India, Bangladesh, other South Asian markets
Potential to increase farmer income by 15-20%

Demo Story

"Ahmad bhai from Faisalabad hasn't used Kissan AI in 3 days. His phone buzzes: 'Ahmad bhai, aapke ilake mein 2 din mein barish hai — aaj hi DAP spray kar lo. Pehle aap ne November mein wheat lagayi thi, ab growth stage pe hai.' He opens the app, confused about a yellow spot on his wheat leaf. He snaps a photo. Qwen Vision analyzes it. He asks in Punjabi: 'Rate kya hai?' The market agent checks Faisalabad mandi: 'Gehu ka rate 2,400 rupay man hai.' Three agents, one memory, one farmer, one decision."