An-SmartFarm

Planning a smart farm means greenhouse costs, equipment, crop income, land prices, and subsidies are scattered, so it is hard to estimate cost and balance in advance. Handing it to an LLM risks hallucinated figures.

Designed it TAG-style: code does the math, the LLM only explains. Every figure is a lookup into 8 public-data CSVs (greenhouse costs, ICT equipment, crop income, land transactions, subsidies, and more) with no hardcoded prices. Each row carries a source URL and confidence, the system prompt forbids recomputing numbers or hallucinating, and if the AI call fails it falls back to a template report and says so on screen.

Produces a plot layout with construction cost, operating balance, and subsidy scenarios, and can cite a public-data source for every number on screen. Honorable Mention at the Honam IS Hackathon.