Skip to content

capblack222/AgriSense-Agent

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 

Repository files navigation

🌾 AgriSense Agent

Turning Weather Data Into Actionable Farm Decisions

A personalized, explainable AI agent helping smallholder farmers make smarter daily choices.

📌 Overview

AgriSense is my capstone project from the Google x Kaggle AI Agents Intensive (2025) - a 5-day deep dive into agentic system design, memory, tools, and multi-agent workflows.

This project holds deep meaning for me. Growing up, I watched farmers in my ancestral village in Kanpur struggle with unpredictable weather and limited access to timely advice. Those experiences shaped my long-term goal of building tech that empowers rural communities.

AgriSense is designed for exactly that purpose.
It converts raw weather forecasts into clear, crop-specific, stage-aware farm recommendations using AI agents.

🌟 Key Features

🤖 Interactive Farm Assistant

  • Asks intuitive questions
  • Fetches real-time weather
  • Generates daily personalized guidance
  • Remembers previous sessions

🌦️ Weather Intelligence

Powered by the Open-Meteo API.

🧠 Context + Memory

Stores crop, stage, past alerts, and weather snapshots.

🪪 Transparent Rule Engine

Agronomy-inspired decision logic for irrigation, heat stress, and pest risk.

🧩 LLM Reasoning Layer

Combines weather + rules + memory → simple, actionable advice.

🧭 End-to-End Orchestration

Handles workflow, user input, memory updates, and logging.


🏛️ Architecture

User
  ↓
Orchestrator Agent
  ↓
Weather Tool → Rule Engine → Memory Store
  ↓
LLM Reasoner
  ↓
Output Layer (JSON + Natural Summary)

🗣️ Example Interaction

Conversation-style flow with crop → location → stage → personalized plan.
Includes weather fetch, irrigation decisions, fungal/pest alerts, and history summaries.


🔧 Tools & Technologies

  • Google AI Studio (Gemini)
  • ADK (Agent Development Kit)
  • Python
  • Open-Meteo API
  • JSON output logs
  • Notebook-based memory system

🚀 Probable Future Extensions

  • IoT soil sensors
  • Voice + local language interface
  • Seasonal planning
  • WhatsApp/Android integration
  • Pest prediction models

🏅 Google AI Agents Intensive

✔ Built a full agent workflow
✔ Tools + memory + orchestration
✔ Multi-agent concepts
✔ Evaluation + observability
✔ Capstone submitted + Kaggle badge


📁 Repository Structure

notebooks/
capstone/
screenshots/
README.md

🔗 Links

Kaggle Notebook:
https://www.kaggle.com/code/capblack222/agrisense-agent


💛 Personal Note

AgriSense is a small but meaningful step toward using technology for rural impact — inspired by the farmers in my village whose resilience shaped my journey.

To read in-depth about it, please go to Kaggle Writeup

About

AgriSense is an AI-powered farm advisory agent that transforms real-time weather data into clear, crop-specific decisions for smallholder farmers. Inspired by the challenges I witnessed in my own village, it blends rule-based agronomy, memory, tool use, and LLM reasoning to deliver personalized, actionable daily guidance.

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors