ExtensionBot is the Extension Foundation’s purpose-built AI platform for the Cooperative Extension System — designed to deliver trustworthy, research-based answers to the public without the high hallucination risk and generic output that characterizes general-purpose AI. Developed in partnership with AI development firm Thia AI, ExtensionBot constrains its responses to curated content from Cooperative Extension institutions, Ask Extension and other research-based entities, grounding every answer in cited sources that users can verify in real time.
This architecture directly addresses the reliability and trust gap that makes general AI less suitable for high-stakes Extension topics — agricultural production, food safety, and community and youth development — where confident but inaccurate output carries real consequences. The platform is model-agnostic by design, allowing the underlying LLM to be updated or replaced as technology and performance requirements evolve.
The platform scales up through MERLIN, the Extension Foundation’s national data management infrastructure, which aggregates and continuously updates content across the Cooperative Extension System. Every contributing institution strengthens the model’s depth and geographic coverage. Rather than one-time data transfers, MERLIN is built for sustained data integrity: institutions can contribute through live API endpoints updated frequently, ensuring ExtensionBot’s knowledge base reflects current Extension research and guidance. As the network of contributing institutions grows, so does the platform’s capacity to serve as a reliable national resource for the public and a demonstration of Extension’s relevance in an AI-driven information environment.
