Please sign up or log in to continue.

AgriSmart Brasil – Multi-Agent AI for Smart Farming

Smart recommendations, daily insights, and AI-powered diagnostics for Brazilian agriculture—built on Google Cloud Run with Gemini AI, FastAPI, and React.

Comment 2

Inspiration

Brazil is a world leader in agriculture, but most producers still lack access to advanced AI tools for daily decision-making and effective farm management. The vision behind AgriSmart Brasil was to empower farmers with cutting-edge technology—turning raw data into real insights that impact both productivity and sustainability.

What it does

AgriSmart Brasil delivers a cloud-based platform with an intuitive dashboard showing real-time farm data, daily insights with the "Generate Briefing" button, management of active crops, direct chat with AI agents for agricultural questions, and visual crop analysis using image uploads powered by Gemini AI.

How we built it

  • Frontend in React & Vite, deployed on Google Cloud Run
  • Backend in Python FastAPI, orchestrating multiple agents and integrating Gemini AI
  • Optional Firestore database for historical analytics and data persistence
  • Well-documented REST APIs, with live endpoints and a public GitHub codebase

We prioritized clean code, modular systems, and clear data flows, so the platform can scale and adapt as real growers' needs evolve.

Diagram

Challenges we ran into

  • Dealing with CORS issues required precise debugging for seamless frontend-backend communication
  • Optimizing image upload and analysis for a smooth real-time experience was a true test
  • Implementing multi-agent chat logic and robust API documentation demanded continuous iteration

Accomplishments that we're proud of

  • Achieved a stable, user-focused MVP ready for hackathon judging
  • Successfully integrated Google Gemini AI for chat and image analysis
  • Designed a clear and modern dashboard focused on farmer usability
  • Deployed the full system serverlessly with Cloud Run and made all endpoints/documentation public

What we learned

We improved our expertise in orchestrating multi-agent systems for real-world scenarios, integrated cutting-edge AI, advanced our cloud development skills, and gained experience with modern APIs and open-source tooling.

What's next for AgriSmart Brasil – Multi-Agent AI for Smart Farming

  • Integrating real farm data sources and IoT devices for live input
  • Expanding the multi-agent system to support additional crops, languages, and regions
  • Enabling farmer collaboration and data sharing features
  • Continuous improvements based on real user feedback and field trials
Share this project:

Updates