Overview

We are working with the ChicoryLane conservation reserve’s website and its data, to map the content into a dynamic document system.

The system is a node/graph based structure, with nodes that may contain pages of mixed media content accessible through a web browser. Nodes will be recursive, containing other node structures as content.

The goal is to extend the previous work, which implemented a RAG-based chatbot, and extending it to an AI platform with agentic capabilities.

Planned innovations include:

  • An automatic feature to incrementally feed data into the existing LLM system
  • Exploring techniques for incorporating relational database information (like ecological attributes concerning plants in the reserve) into the knowledge base
  • Architectural and cost-based evaluation of development options for a mainstream vertical integration stack

Contacting The Team

You can contact the team by emailing our shared mailbox at:

cichorum@unc.edu.

A member of the team will respond to you, or will defer to another member that might be better suited to handle your request.

Link to original

Initial Project Proposal

The original project proposal can be found here at #6. The provided specifications document is below.