CLARION: Clarifying Language through Agentic Refinement with Iterative Operational Navigation

CLARION (Clarifying Language through Agentic Refinement with Iterative Operational Navigation) is a human-in-the-loop, multi-agent AI system designed to reduce ambiguity in natural language guidance through iterative refinement. Built in Python with coordinated OpenAI agents and a Flask-based HTML interface, the system structures clarification into specialized roles that surface assumptions and generate targeted questions for the user. The prototype demonstrates how coordinated agents can systematically resolve uncertainty and improve alignment to enable robust AI-aided decision-making in high-stakes domains.

Intern: Gail McKinney

Mentor: Samantha Levy (AOS)