AI-Powered Multi-Agent System for Smart Crop Recommendation Problem Statement
India faces massive post-harvest losses — 15–20% of cereals and up to 30% of perishables are lost every year due to improper crop selection, storage, and handling. Farmers often struggle to decide which crops are best suited for their soil and climate, leading to inefficiencies across the supply chain.
Our Solution
We built an AI-powered multi-agent system that integrates soil and climate data to provide end-to-end intelligence for farmers. The current version focuses on one specialized agent:
Crop Recommendation Agent → analyzes soil and climatic parameters (pH, organic matter, rainfall, temperature) and recommends the most suitable crops for sustainable farming.
By aligning crop selection with soil-climate compatibility, the system reduces risks of crop failure, enhances productivity, and indirectly minimizes post-harvest losses.
Profitability ∝𝑓 ( Soil compatibility , Climate suitability)
Tech Stack
AI Models: IBM Granite models Framework: IBM Agent Development Kit (ADK) Data: Agricultural datasets (soil, climate, crop suitability) Evaluation: Model outputs cross-verified with agronomy research guidelines
What We Learned
How AI can be adapted to agriculture beyond buzzwords. The importance of structured multi-agent design in building scalable systems. Data quality is the true bottleneck — without reliable agricultural data, AI has limited impact.
Challenges
Limited access to high-quality, localized datasets. Balancing accuracy across diverse regions of India. Making AI insights usable and trustworthy for farmers with varying digital literacy.
Future Scope
Add new agents for storage monitoring, pest management, and supply chain optimization. Build a farmer-facing app with local language support. Scale the system for regional food distribution and market linkage.
Impact
This project provides a scalable pathway to: Reduce crop mismatch and post-harvest losses Improve farmer profitability Strengthen food security and sustainability
Built With
- and-integration-with-ibm-granite-models.-javascript-(node.js)-?-if-you?re-making-a-web/app-interface-for-farmers-or-operators.-yaml/json-?-for-agent-configuration-in-the-ibm-agent-development-kit-(adk).-sql-?-for-handling-soil
- climate
- data-preprocessing
- javascript
- python
- sql
- yaml/json
