AgriAI

AgriAI: From Soil to Sale — Smart Farming. Smarter Future. Empowering farmers with AI, IoT, and real-time insights, Web3-inspired platform that grows with you from crop choice to market success.

Comment

Inspiration

Farming communities in rural areas often lack easy access to agronomic advice. Inspired by conversations with local farmers and the challenge theme of digital inclusion, we envisioned a solution that could talk to farmers in their language, provide timely AI-driven guidance, and eventually integrate with IoT-based field monitoring.

What it does

🌾 AgriAI: Smart Agricultural Assistant Powered by AWS is a voice-enabled and web-based AI agent that assists farmers with critical decisions across the agriculture lifecycle—from crop selection to pest detection, yield prediction, and even market selling insights. Built using AWS generative AI services (Amazon Bedrock & SageMaker) and next-gen connectivity, it aims to reduce the information gap for small-scale farmers by providing intelligent, localized, and real-time support.

How we built it

AI Assistant: Integrated Amazon Bedrock (Claude 3 Sonnet) to provide human-like responses on topics like crop selection, pest remedies, and pricing strategy.

** ML Models on SageMaker:**

crop-recommendation-endpoint: Suggests optimal crops based on soil and weather.

pest-detection-endpoint: Accepts image input (simulated) and predicts pest issues.

yield-prediction-endpoint: Forecasts harvest quantity using historical inputs.

AWS Cognito: Enabled secure access without forcing full login, great for mobile access.

Voice Support (Optional): Used Amazon Polly + Transcribe to make it voice-accessible.

Frontend: Built using React + Vite with clean UX for farmers and dashboard for agri experts.

Challenges we ran into

Cost optimization: Managing SageMaker and Bedrock API usage while staying within Free Tier.

Latency in voice agent: Real-time transcription and response timing needed careful tuning.

Localization: Supporting multilingual queries for broader farmer reach is an ongoing effort.

Security: Safely calling AWS APIs from frontend required strict access rules and Cognito setup.

Accomplishments that we're proud of

✅ Built a fully functional AI assistant that can speak, understand, and respond like a human — helping farmers or customers without human intervention.

🚀 Integrated AWS Bedrock and SageMaker successfully to deliver real-time, intelligent responses and ML predictions using the Free Tier efficiently.

📞 Enabled voice interaction using Amazon Polly and Transcribe — allowing users to talk to the AI instead of typing, making it accessible to low-literacy users.

🌱 Created a smart crop advisor and pest detector with minimal latency using SageMaker endpoints trained on real-world datasets.

🔐 Implemented secure access with Cognito Identity Pools to allow usage without forcing full signups or logins.

📊 Developed a clean, intuitive frontend dashboard for farmers, business owners, and admins to monitor conversations, bookings, and analytics.

🛠️ Overcame API latency and authentication issues to deliver a smooth user experience despite working across multiple AWS services.

🧠 Crafted context-aware AI prompts that allow the system to remember the conversation and react appropriately — making it feel like a real assistant.

What we learned

How to build and deploy custom endpoints in SageMaker efficiently

The power and simplicity of Amazon Bedrock for real-world generative AI use cases

Integrating multiple AWS services (Cognito, Bedrock, SageMaker, Polly) under a unified architecture

Techniques to simulate IoT sensor data and use AI to interpret it

Built With

Share this project:

Updates