🌾 NobinKrishi - From Soil to Sale
The Story That Inspired Me
In Mithapukur, Rangpur, a farmer named Md Mohsin Ali cultivated 12,600 kilograms of potatoes on just 1.5 acres of land.
After months of work, he sold his harvest for just Tk 7.50 per kilogram because he had nowhere to store it. Meanwhile, potatoes were selling for Tk 32–35 per kilogram in Dhaka.
He earned Tk 107,800 from his harvest, but spent approximately Tk 75,000 on cultivation, seeds, fertilizers, pesticides, and harvesting. Once land expenses and his own labour were considered, he expected to make a loss.
I came across his story while researching the challenges faced by farmers in Bangladesh. What struck me was the contradiction: a farmer could produce thousands of kilograms of food and still struggle to earn a reasonable living from it, while consumers elsewhere paid several times more.
The problem wasn't simply agricultural productivity. It was the lack of access to information, storage, market connections, and opportunities.
That story stayed with me.
Agriculture feeds Bangladesh, yet the people who work hardest to produce our food often have the least access to information, technology, and fair markets.
A farmer can spend months growing a crop only to discover that market prices have fallen. Another might lose an entire harvest because a crop disease was identified too late. Many farmers still depend on intermediaries to sell their produce, while consumers pay significantly more for the same food.
What if farmers had access to the kind of information and technology that could help them make better decisions, without needing expensive equipment or technical expertise?
That question inspired NobinKrishi (নবীন কৃষি).
What I Built
NobinKrishi is an AI-powered agricultural platform designed to make farming smarter, markets fairer, and agricultural knowledge more accessible.
I wanted to create something that could support a farmer throughout the entire farming journey, rather than solve just one isolated problem.
The platform brings together:
- Smart Marketplace: Connects farmers directly with consumers and buyers, helping reduce unnecessary intermediaries and food waste. It also allows consumers to discover affordable, imperfect produce.
- AI Land Analysis: Helps farmers identify suitable crops based on soil conditions, location, climate, and water availability.
- Crop Disease Scanner: Uses image-based AI to analyze crop photographs and provide potential disease identification and treatment guidance.
- Bangla Voice AI Assistant: Allows farmers to ask agricultural questions in their own language without having to navigate complicated text-based interfaces.
- Demand Forecasting: Helps farmers understand market trends and anticipate which crops may have higher demand.
- Smart Weather Intelligence: Provides localized weather forecasts, agricultural recommendations, and alerts.
- Crop Knowledge Grid: Brings growing guides, disease information, crop requirements, and market insights into one accessible place.
- Personal Farm Dashboard: Provides a centralized space for monitoring farm activities, expenses, crop health, and harvest planning.
My central idea was to connect information that is usually scattered across different sources into one accessible agricultural companion.
How I Built It
I developed NobinKrishi independently, taking responsibility for the concept, research, product design, development, and integration of its different components.
I approached the project as an interconnected ecosystem.
1. AI and Machine Learning
I explored language models, conversational AI, and computer vision to make agricultural assistance more accessible. My technology exploration included:
- Dhenu Llama 3 for Bangla-language interactions
- Vision-based crop analysis
- GPT-based reasoning
- Gemini for multimodal AI capabilities
- Whisper for speech recognition
2. Agricultural and Environmental Data
I explored agricultural and geospatial sources, including:
- BARC, BRRI, and BARI
- FAO
- SoilGrids
- NASA POWER
- Sentinel-2
- Landsat imagery
3. Platform Development
I worked on the web experience, marketplace, farm dashboard, agricultural information systems, and the integration of AI-powered features into a unified interface.
4. Accessibility
One of my biggest priorities was making the platform understandable to people who might not be comfortable with complicated technology. This influenced my decisions around Bangla voice interaction, visual interfaces, and simplified navigation.
Challenges I Faced
Building an agricultural ecosystem alone was considerably more challenging than building a conventional web application.
The biggest challenge was scope. Agriculture involves interconnected problems: unpredictable weather, crop diseases, soil conditions, market uncertainty, supply-chain inefficiencies, and food waste. I had to think about how these different systems could work together instead of treating them as separate features.
Another challenge was accessibility. An advanced AI assistant is not particularly useful if the intended user cannot comfortably communicate with it. This pushed me to explore voice-based interactions and localized agricultural assistance.
I also had to consider the reliability of agricultural recommendations. AI-generated advice must be grounded in relevant agricultural information rather than simply sounding convincing.
Working independently meant constantly balancing research, development, design, and problem-solving.
What I Learned
This project taught me that meaningful AI innovation is not about adding artificial intelligence to everything.
It is about identifying where intelligence, information, and accessibility can genuinely improve people's lives.
I learned to think about technology as an ecosystem, explore the importance of localized AI, work with different categories of data, and approach product development from the perspective of its intended users.
Most importantly, I learned that agricultural innovation should not only ask how we can grow more food. It should also ask how we can make sure the people growing that food have access to better opportunities.
What's Next?
I envision NobinKrishi becoming a reliable digital companion for smallholder farmers across Bangladesh and, eventually, other agricultural communities.
I want to continue improving its agricultural intelligence, validating its recommendations with real-world data, expanding Bangla and regional-language support, and exploring partnerships that can bring the platform closer to farming communities.
The farmer in Rangpur who sold thousands of kilograms of potatoes without receiving a fair return represents a much larger problem.
I cannot solve every challenge in agriculture with one application. But I can build something that works toward changing how farmers access information, opportunities, and markets.
🌱 The Vision
From soil to sale, every farmer deserves a fairer opportunity.
That's the future I'm building with NobinKrishi.
Built With
- agtech
- ai
- climate
- genai
- llms
- machine-learning
- natural-language-processing
- react
- typescript
