# Three-Minute Video Pitch Script

No slides. Founder on camera. Show the demo or research assets briefly on
screen if possible.

## 0:00-0:25 | Team

Hi, I am Yifan Yang, founder and technical lead of Rayford AI. I work with
technical and scientific guidance from Dr. Lei Zou, Dr. Zhengzhong Tu, and Dr.
Heng Cai. My work focuses on disaster resilience, street-view damage assessment,
visual-language models, and autonomous GeoAI. I have built public research code
and datasets for post-disaster property damage assessment, including
bi-temporal street-view analysis and multimodal damage arbitration.

## 0:25-0:55 | Problem

After hurricanes, floods, severe storms, and wildfires, insurers and local
governments need to know which properties were damaged, how severe the damage
is, and what evidence supports each decision. Today that process is still slow
and fragmented. It depends on field visits, manual photos, forms, and repeated
review. The result is delayed claims, delayed aid, and poor prioritization when
communities need speed.

## 0:55-1:35 | Solution

Rayford AI builds Ray, a resilience AI assistant for every property. Our first
product workflows are Ray Assess and Ray Claims. They fuse pre-event and
post-event street-view, satellite, drone, and field imagery with property data
to create damage scores, confidence estimates, and auditable evidence packages
for insurers, adjusters, and local governments.

We are not trying to replace adjusters first. We help them triage faster, see
the evidence clearly, and focus human review where it matters most.

## 1:35-2:10 | What We Built

The technical foundation already exists in our research. We have built
bi-temporal street-view disaster assessment, visual-language disaster damage
perception, CLIP-enhanced multimodal arbitration, and satellite-to-street
generation. In April 2026, Mosaic featured our Texas A&M Hurricane Milton
research and described why street-level imagery is needed to complement
satellite and aerial disaster assessment. The commercial MVP turns these
research assets into one workflow: select a disaster area, link imagery to
properties, estimate likely damage, explain the evidence, and export a claims
or inspection report.

## 2:10-2:40 | Why This Team

This market needs a team that understands geospatial AI, disaster resilience,
computer vision, and the built environment. Rayford AI is starting from years of
focused research, not a generic AI wrapper. We understand the data, the models,
the limits, and the need for human trust in disaster decisions.

## 2:40-3:00 | AggieX Goal

During AggieX, we will complete 40 customer interviews, build a Ray Assess web
demo, validate two historical disaster case studies, define a paid pilot offer,
and secure three serious pilot or LOI conversations. Our long-term vision is to
make every exposed property ready, assessable, and recoverable.