AI Assistant for Resource Directories

Conversational AI + RAG· Feb. 2025–Present· Columbus, Ohio

Callie is a prototype conversational AI assistant that helps families navigate community resource directories for food, housing, and healthcare, using natural language and multilingual support, with answers grounded in vetted directory data. I led product strategy and pitched the concept at Bold Crow, drawing on personal experience and years of volunteering to identify the underlying problem.

Challenge / Problem

Families often abandon community resource directories after failed searches or confusing terminology. Information is frequently outdated or incomplete, leading to poor matches and eroded trust, and organizations lack visibility into what families are actually looking for.

What We Did

Through user discovery, we prototyped Callie: an assistant that understands natural language requests instead of requiring the "right" search terms, offers 24/7 multilingual support, uses Retrieval Augmented Generation (RAG) to ground answers in vetted resource data, and integrates with Model Context Protocol (MCP) for real-time updates. Built with Python, OpenAI API, Google Cloud Storage, and Pinecone.

Impact / Insights (Pre-Launch)

Conversational AI shows strong potential to reduce directory search abandonment
Early prototype feedback pointed to meaningful time savings for both families and staff
Nonprofits could gain visibility into trending needs and service gaps — insight that could also strengthen grant-writing and funding pitches

Bold Crow case study →

Written from Bold Crow's perspective, not mine.