AgroRithm

AgroRithm uses AI, IoT, geofencing, SMS, BLE, computer vision and local voice AI to prevent farmer-herder clashes, reward eco-actions with tokens, and work offline via PWA app, USSD, Telegrambot etc

Comment

Inspiration

Rural communities across Nigeria and Sub-Saharan Africa face persistent farmer-herder clashes that have escalated into one of the region's most pressing security challenges. These conflicts, intensified by climate change, desertification, and population pressure, result in thousands of deaths annually and the displacement of entire communities. The traditional grazing routes have been disrupted by expanding farmlands, leading to resource competition that destroys livelihoods and threatens food security. We were inspired by stories from communities in Plateau, Benue, and Kaduna states, where families have lost everything to these conflicts. During field visits, we witnessed the devastating impact on both farming and herding communities - children unable to attend school, families displaced from ancestral lands, and entire communities living in fear. The challenge isn't just about technology - it's about building bridges between communities that have coexisted for centuries.

What it does

Real-time Conflict Prevention: Detects cattle intrusions or strays using either IoT, geolocation, BLE beacons attached to livestock and AI-powered computer vision, OR the PWA(Progressive Web App or mobile app or telegram bot. It implements geofencing around farmlands with customizable alert zones and then sends immediate alerts via SMS, Telegram bot, in Progressive WebApp or mobile to prevent conflicts before they escalate. Provides early warning systems for seasonal migration patterns, The core of this product works fully offline.

Blockchain-Based Traceability and Sustainability Incentive System:

Rewards peaceful conflict resolution and sustainable practices with Hedera and TON tokens(testnet). Creates an NFT-based cattle registration system for livestock tracking and ownership verification. Implements smart contracts for automated reward distribution based on community-verified actions. Establishes a circular economy where tokens can be redeemed for agricultural inputs and tools, and also sold to international companies by partnering with the Ministry of Environment, Nigeria, West African Alliance on Carbon markets and Finance and the Global Carbon Council. Offline-First AI Assistant: Deploys TinyLLaMA + Langchain for multilingual agricultural advice that works without internet Provides voice interaction through Coqui TTS and Vosk STT in Hausa, Pidgin. Offers context-aware recommendations for crop management, livestock care, and conflict mediation. Maintains a knowledge base of local farming practices and traditional conflict resolution methods Multi-Platform Accessibility: Mobile app (React Native/ PWA/ Telegram bot/USSD) for farmers and herders with live tracking and IoT integration/BLE scanning, Telegram bot integration for communities with limited smartphone access, Voice-based interface for low-literacy users, Works across feature phones, smartphones, and basic internet connections.

How we built it

We built: A mobile app (React Native + Firebase) and web app (ReactJS) for farmers/herders with live cow tracking and monitoring to prevent stray or theft, also using BLE scanning, and voice support. A Telegram bot that uses NFTs for cow records, sustainability actions, and AI help. An AI model using YOLOv5 to detect cows from a laptop camera or an IoT device. TinyLLaMA + Langchain for multilingual AI advice offline. Coqui TTS + Vosk STT for voice communication. Integration with Hedera for rewards, traceability and Google Firebase for logs. Use of SMS alerts for internet constrained areas, USSD to access services in low internet areas, etc.

Challenges we ran into

Making all systems work offline or in low-bandwidth areas. Deploying tinyllama to a low resource phone was challenging due to low RAM and ROM space and also tech dependencies. Technology for AI models to voice local languages. BLE hardware limitations on older Android phones Tokenising local actions meaningfully for sustainability rewards. Balancing security with accessibility. Dependency challenges in building on react native. Integrating the various systems.

Accomplishments that we're proud of

Functional Telegram bot with TON NFTs and TinyLLaMA support totally offline. Voice support in Hausa, Pidgin. BLE + AI-based real-time cow detection with alerts. Reward system for herders/farmers using Hedera and Firebase. Offline AI chatbot trained on local agriculture knowledge. Building a robust solution that's offline and syncs when there is network. Blockchain traceability and reward system based on tech and community verification. logging data to local dB when offline, syncing when online. Log tokens even when offline. Use of gps without network, cached map via PWA caching when fully offline. Building a community wide chat for admin, security, herders and farmers for realtime intelligence sharing even offline.

What we learned

How to run AI tiny models and voice tech offline using limited hardware. Real-world impact of peace-tech in agriculture. Firebase Persistence to send chats, tokens offline, and they go when there is network, it syncs. Designing for low-literate, multilingual, rural communities. Blockchain tokenomics for behaviour change. Community adoption depends more on user experience design than technological sophistication. Cost Optimisation:- Creating a cost-efficient and portable device for rural dwellers. Designing an almost fully offline tech integrating web2 and web3, a good mix of current and emerging tech, a good mix of 3rd and 4th industrial revolution because of the fine mix of IoT + Software + data connectivity.

What's next for AgroRithm

Immediate Expansion (3-6 months): Scale pilot program to 50 communities across Nigeria's Middle Belt Deploy 1,000 IoT kits to participating farmers and herders Launch a comprehensive training program for community facilitators Establish mobile money integration for token redemption

Technology Development (6-12 months): Expand AI agent capabilities with satellite data integration for conflict hotspot mapping Develop drought and climate prediction models using machine learning Create a marketplace platform for agricultural products and services Regional Scaling (1-2 years): Expand to Chad, Niger, and Cameroon with localised adaptations Establish partnerships with regional agricultural organisations Create policy frameworks for technology adoption at the governmental level Develop train-the-trainer programs for sustainable knowledge transfer Long-term Vision (2-5 years): Become the leading peace-technology platform for pastoral communities across Africa Establish AgroRithm as a model for climate adaptation and conflict prevention Create sustainable revenue streams through carbon credit programs and impact investing Influence policy development around technology-enabled conflict resolution

Built With

Share this project:

Updates