Farmer Advisory Agent with Weather and Crop Knowledge Tools

Can a small local language model give useful farm advice if it has real tools and remembers the farmer? This project builds a smolagents tool-calling advisor with a crop knowledge tool, a weather tool on stored Open- Meteo forecasts, a crop calendar, and Mem0 long-term memory per farmer. The model (qwen2.5:3b) runs in local Ollama on a laptop with RTX 2050 4 GB (Ryzen 7, 8 GB RAM).

NOTE: Decision support for teaching and research only, not agronomic or pesticide advice. A small local model can be wrong and the forecasts are stored ones: check with your local agriculture extension officer and follow the product label. No doses are given by this system. The crop calendar is a teaching table written from general knowledge, not checked against FAO or state sources.

Category: Department of Computer Science and Engineering / Agent Development

Agent Development Project AG18 from ProTechSC Academy, with graphical abstract, video demonstration and project guidance.