Here is a complete draft for your hackathon or project submission, integrating the technical details from your slides and tailored to your development stack:
Inspiration While exploring solutions for the agricultural sector, we noticed a glaring gap: most modern agritech relies heavily on consistent, high-speed internet, leaving farmers in remote, low-connectivity areas behind. Farmers continuously struggle with delayed crop disease diagnosis, exploitative middlemen eating into their profits, and navigating complex portals for government subsidies. We wanted to build a comprehensive, offline-capable ecosystem that truly empowers the farmer—providing instant help in the field, fair market pricing, and seamless financial security.
What it does AgriNexus AI is a comprehensive 6-module ecosystem designed to be a farmer's "Doctor, Manager, and Broker":
Offline AI Disease Detection: Diagnoses crop issues instantly without the internet using on-device AI, providing prescriptions with 95% accuracy.
AI Quality Grading: Evaluates crop quality (Grade A/B/C) with 98% accuracy to ensure transparent and fair pricing.
Direct Selling via ONDC: Connects farmers directly to a pan-India network of buyers, cutting out the middleman entirely.
Secure Escrow Payments: Protects transactions with middleman-free, direct-to-bank escrow transfers.
Equipment Rental Marketplace: Allows farmers to book tractors and drones with real-time GPS tracking.
Digital Vault + Schemes: Automatically checks eligibility and auto-fills applications for over 100 government agricultural schemes.
How we built it We developed a robust full-stack architecture to handle both the offline and online demands of the platform. The backend infrastructure is powered by Node.js and hosted on Google Cloud to ensure scalable, reliable performance. To achieve zero-connectivity diagnosis, we trained and compressed Convolutional Neural Networks (CNNs) using TensorFlow Lite, allowing them to run locally on mobile devices. We also deeply integrated the Google Gemini API to parse complex government schemes and power the intelligent auto-eligibility engine. The frontend was built using HTML, CSS, and JavaScript, focusing on an intuitive, accessible user experience.
Challenges we ran into The most significant hurdle was the "model compression" phase—optimizing our AI disease detection and grading models to run natively on low-end smartphones without sacrificing our target accuracy of 95%+. Additionally, integrating the ONDC protocol for the direct-selling marketplace required handling complex, asynchronous state management to ensure the secure escrow payment flow remained airtight and middleman-free.
Accomplishments that we're proud of We are incredibly proud to have completed a fully functional MVP that establishes the complete "Doctor-Manager-Broker" loop. Successfully deploying a TensorFlow Lite model that actually works accurately in a zero-connectivity environment is a massive win for our team. Furthermore, integrating Gemini to turn the nightmare of bureaucratic government scheme forms into a seamless, one-click application process feels like a true leap forward for rural tech accessibility.
What we learned Building AgriNexus AI pushed us to master on-device machine learning and edge computing. We gained deep, practical experience working with the ONDC network and understanding decentralized e-commerce protocols. We also learned how to effectively leverage Large Language Models like Gemini not just for chatbots, but for parsing and automating complex, real-world administrative tasks.
What's next for AGRI NEXUS AI Our development timeline over the next 12+ months is focused on scaling impact and accessibility:
1-3 Months: Launching IoT soil sensor pilots, improving offline data packs, and further compressing our ML models.
3-6 Months: Expanding our ML capabilities to include predictive price forecasting and deploying a fully multilingual UX.
6-12 Months: Transitioning to a Voice-first UI to eliminate literacy barriers, expanding regional languages, and integrating drone technology.
12+ Months: Launching a dedicated livestock health module, scaling the platform state-wide, and forging official partnerships with Farmer Producer Organizations (FPOs).
Built With
- css3
- geminiapi
- html5
- javascript
- next.js
- node.js
Log in or sign up for Devpost to join the conversation.