Back to Projects
Side project

Zafra

Agentic AI for agricultural management and optimization

Zafra

Problem

Producers in Northwest Argentina need faster decisions with data scattered across livestock, crops, and weather.

Context

Zafra was created to bring applied AI to the field in northern Argentina. Instead of rigid dashboards, it offers an ecosystem of agents that interpret territorial data, coordinate specialized sub-agents, and return actionable answers to producers and technical teams.

Key features

  • Agentic architecture with sub-agents for agriculture and livestock in one flow.
  • Satellite data integration for monitoring and early detection of critical events.
  • Configurable alerts based on business thresholds and field conditions.
  • Conversational assistance via Telegram for low-friction queries and follow-up.
  • Designed around real use cases in Northwest Argentina, not lab-only prototypes.

My contribution

Designed and implemented the agentic system: sub-agent orchestration, satellite integration, and Telegram channel.

Outcome

Zafra strengthens my AgTech and applied AI work: a product built for the territory, connecting satellite insight with daily decisions and broader access to intelligent tools for agriculture.

Stack

PythonAgentic AISub-agentsSatellite dataTelegramLivestockAgricultureAgTech