Name
Communicating CRS and Flood Risk Through Agentic AI: A New Paradigm of Community Engagement and Decision Support
Date & Time
Thursday, May 7, 2026, 9:30 AM - 9:45 AM
Description

Gulf Coast communities face persistent challenges in understanding flood risk and navigating FEMA’s Community Rating System [CRS), particularly when staff capacity is limited and scientific information is difficult to translate into everyday decisions. These barriers limit effective community engagement and slow progress toward flood-safe, resilient communities. To address these communication challenges, we developed an agentic AI-driven CRS Communication Dashboard that integrates flood hazard data, NFIP Risk Rating 2.0 insights, community socioeconomic indicators, and local mitigation information into an interactive, easy-to-use environment for Gulf Coast communities.

The dashboard introduces a new paradigm for science communication and community engagement. An autonomous AI assistant helps users interpret technical terms, generates plain-language explanations of CRS activities, prepares localized community briefs, supports conversations with residents, elected officials, and planning staff, and highlights how CRS activities strengthen community resilience. Through digital-twin visualization, interactive flood-scenario exploration, and localized community briefs, the system bridges the gap between complex hazard science and practical resilience action and helps communities clearly see the connection between flood risk, mitigation actions, and potential insurance benefits.

Co-developed with Louisiana stakeholders through an extension and outreach framework, this approach enhances community engagement around CRS, improves communication of flood insurance and mitigation benefits, and supports the development of more flood-safe, resilient Gulf Coast communities. It offers a scalable model for strengthening engagement, building local capacity, and promoting resilience across the Gulf Coast region.

Location Name
201D
Is presenter a student?
No