Agri Expert – AI-Powered Agricultural Intelligence for Nepal

Smart farming made simple: Diagnose diseases, predict markets, and get climate-smart advice—anytime, anywhere.

Comment

Inspiration

Agriculture is the backbone of Nepal, yet the farmers who feed the nation remain the most underserved when it comes to technology and information access. During research, I was shocked to learn that diseases like Late Blight and Bacterial Wilt destroy up to 40% of crop yield every year in Gandaki Province—often because farmers receive diagnosis too late.

I also saw how unpredictable weather, unreliable traditional calendars, and unstable market prices leave farmers financially vulnerable.

This inspired me to build Agri Expert—a tool that puts AI-powered agricultural intelligence directly into the hands of rural farmers, even if they don’t have stable internet. My goal was simple: to build something that genuinely protects farmers, improves their income, and supports sustainable agriculture in Nepal.

What it does

Agri Expert is a mobile-first, AI-powered companion for farmers. It delivers:

🦠 AI Crop Doctor (Vision)

Upload a crop photo → get instant disease detection using Google Gemini 2.5

Shows disease cause, severity %, treatment, and prevention

Save Analysis Card generates a digital pamphlet that works offline

📈 Market Mandi (AI Intelligence)

Visualizes crop price trends

AI Market Advisor predicts whether to “Sell Now” or “Wait”

Offline fallback using LocalStorage

🌤️ Climate + Culture Advisory

Real-time weather

Indigenous Nepali Calendar

Altitude-based recommendations

🗣️ Inclusive Design

Audio assistant for low-literacy users

Simple icons

Offline-first UI

In short: Diagnosis + Climate + Market = One powerful AI tool for Nepali farmers.

How we built it

The entire project was built from scratch during the SNOW FEST Hackathon 2025.

Frontend:

React + TypeScript

Tailwind CSS

Recharts for visualization

Lucide Icons

AI Engine:

Google Gemini 2.5 Flash (Vision + Text)

Custom prompts for disease detection, market reasoning, and climate guidance

Offline Tech:

LocalStorage caching

HTML2Canvas for Save Card

Architecture:

Modular components

Service-based API handlers

Optimized mobile-first UI

I followed an iterative approach:

Built the UI skeleton

Integrated Gemini Vision

Implemented Save Card feature

Added market intelligence

Designed climate + cultural logic

Tested everything for low-connectivity environments

Challenges we ran into

Every meaningful project has obstacles—here were mine:

❗ 1. Vision AI accuracy for Nepali crop diseases

Gemini was powerful, but designing prompts to detect local diseases like Bacterial Wilt required experimentation and dataset understanding.

❗ 2. Offline-first functionality

Many rural farmers lack stable internet, so caching, fallback modes, and offline save-cards required careful planning.

❗ 3. Designing for Low-Literacy Users

Using icon-based UI, audio assistance, and simple Nepali-friendly layouts took time but dramatically improved usability.

❗ 4. Market insights complexity

Creating an AI agent that gives “Sell Now / Wait” advice based on short-term and seasonal insights required tuning reasoning prompts.

❗ 5. Hackathon time constraints

Building an AI-powered, multi-module app in a short timeline was challenging but incredibly rewarding.

Accomplishments that we're proud of

🏆 1. Vision AI that works for Nepali crops

Getting accurate disease detection and severity estimation felt like a breakthrough.

🏆 2. Offline Save Analysis Card

This ensures farmers can still use treatment steps even without internet.

🏆 3. Market Advisor that feels truly helpful

It’s simple, but farmers can use it immediately to avoid bad selling decisions.

🏆 4. A clean, accessible UI

Mobile-friendly, intuitive, and usable for rural audiences.

🏆 5. Built entirely during the hackathon

No old code, no shortcuts—everything built fresh, responsibly, and originally.

What we learned

This project taught me:

How to use Google Gemini Vision for real agricultural problems

How to build offline-first web apps

How crucial UX accessibility is for non-technical users

How to design AI reasoning for real-world decisions

How much impact a well-designed tool can have on farmer livelihoods

Most importantly, I learned that technology becomes meaningful only when it uplifts real people.

What’s next for Agri Expert – AI-Powered Agricultural Intelligence for Nepal

Agri Expert is just the beginning. Future improvements:

🔮 Expand to more crops

Tomato, cauliflower, citrus, wheat, millet, and more.

🔊 Full Nepali voice interface

Voice-driven disease diagnosis & recommendations.

🌱 Smart Soil Monitoring

IoT sensors for moisture, pH, and nutrient detection.

🛰️ Drone vision for large-field scanning

High-scale disease mapping.

🧾 Farmer profile + crop diary

Track planting, spraying, and harvest cycles.

🌍 Multi-language support

Nepali, Magar, Gurung, Bhojpuri.

📡 Government partnership mode

Integrate with agricultural offices and agri-extension workers.

Built With

  • analysis
  • api
  • apis
  • as
  • caching)
  • card
  • cloud
  • crop
  • css
  • css3
  • data
  • datasets
  • developer
  • engineering
  • export
  • flash
  • frontend
  • gemini
  • html2canvas
  • html5
  • icons
  • if
  • image)
  • javascript
  • localstorage
  • machine-learning
  • market
  • offline
  • open-meteo
  • optional/add
  • or
  • price
  • prompt
  • public
  • react
  • recharts
  • save
  • tailwind
  • tech
  • tools
  • typescript
  • ui
  • used)
  • vision/reasoning
  • visualization
  • vite
  • weather
  • whichever
  • you
Share this project:

Updates