# AggieX Summer Accelerator Application Draft

This draft is written for the Rayford AI direction. Replace bracketed fields
before submission.

## Section 1 | Team Profile

**Company / Venture Name**  
Rayford AI

**Primary Contact Email**  
[TODO: primary contact email]

**Venture Stage**  
Functional research prototype; active research commercialization; pre-revenue;
MVP under development.

**Industry / Vertical**  
InsurTech, GovTech, Climate Resilience, GeoAI, Disaster Technology, SaaS.

**Website**  
Current research site: https://autogeoai4sci.github.io/  
Company site: https://rayford-ai.com

**Legal Structure & IP**  
Rayford AI is currently pre-incorporation. No company IP filings have been made
yet. The founder is Yifan Yang, with technical and scientific guidance from Dr.
Lei Zou, Dr. Zhengzhong Tu, and Dr. Heng Cai. The technical foundation comes from
related academic research in bi-temporal street-view disaster assessment,
visual-language disaster damage perception, multimodal arbitration, and
satellite-to-street disaster imagery. Before incorporation, pilots, or IP
filings, the team will review university IP, conflict-of-interest, and licensing
requirements through the appropriate Texas A&M commercialization process. The
team will evaluate patentability around auditable multimodal damage assessment,
property-level evidence packaging, and pre-event/post-event resilience
intelligence workflows.

**Video Pitch URL**  
[TODO: unlisted YouTube, Loom, Drive, or Dropbox URL]

## Section 1.1 | Founding Team

**How many founders do you have?**  
1

**Founder #1 First Name**  
Yifan

**Founder #1 Last Name**  
Yang

**Founder #1 Role**  
Founder and Technical Lead

**Founder #1 Email**  
[TODO: founder email]

**Founder #1 LinkedIn URL**  
[TODO: LinkedIn URL]

**Founder #1 Equity %**  
[TODO: confirm company equity structure before submission]

**Founder #1 Background**  
Yifan Yang is a Texas A&M GeoAI researcher working on disaster resilience,
street-view damage assessment, visual-language models, and autonomous GeoAI.
He has built public research code and datasets for post-disaster damage
assessment, including bi-temporal street-view analysis and multimodal damage
arbitration.

**Founder #1 Resume**  
[TODO: upload resume]

**Founder #1 TAMU Affiliation**  
[TODO: PhD student / graduate researcher / other official affiliation]

**Founder #1 College**  
[TODO: College name]

**Founder #1 Major**  
[TODO: Major or program]

**Founder #1 Expected/Actual Graduation Date**  
[TODO: MM/YYYY]

## Section 1.2 | Advisors

**How many advisors do you have?**  
3

**Advisor #1**
Dr. Lei Zou, Scientific and Technical Advisor. Dr. Zou provides guidance on
GeoAI, disaster resilience, damage assessment, and research commercialization.
[TODO: confirm official title, affiliation, email, and permission to list.]

**Advisor #2**
Dr. Zhengzhong Tu, Technical Advisor. Dr. Tu provides technical guidance on
computer vision, AI model design, and validation. [TODO: confirm official title,
affiliation, email, and permission to list.]

**Advisor #3**
Dr. Heng Cai, Technical Advisor. Dr. Cai provides technical guidance on the
built environment, infrastructure resilience, and practical validation for
property-level disaster intelligence. [TODO: confirm official title,
affiliation, email, and permission to list.]

## Section 2 | Problem & Solution

**What problem are you solving, and who experiences it?**  
After hurricanes, floods, severe storms, and wildfires, insurers, adjusters,
local governments, and recovery teams need to know which properties were
damaged, how severe the damage is, and what evidence supports each decision.
Today this work still depends on slow field visits, fragmented photos, manual
forms, and repeated disputes. The cost is time, staffing, claim leakage,
delayed recovery, and poor prioritization. NOAA reported 27 U.S. billion-dollar
weather and climate disasters in 2024 with about $182.7 billion in damage. FEMA
also describes a damage assessment workflow that begins with local collection
and state or tribal verification, which shows how operationally heavy this
process remains.

**What is your solution?**  
Ray is a property-level resilience AI assistant. Our first workflows, Ray Assess
and Ray Claims, fuse pre-event and post-event street-view, satellite, drone, and
field imagery with parcel data to score building damage, explain the evidence,
rank claims or inspections, and export auditable reports for insurers,
adjusters, and local governments.

**Why does your solution win over everything that already exists?**  
Ray is built around before-and-after property evidence, not only hazard maps or
single post-event images. It combines street-level visual reasoning,
satellite-to-street context, multimodal arbitration, confidence scoring, and
human review. The mechanism is an auditable evidence package for each property:
what changed, where it changed, how confident the model is, and which imagery
supports the assessment.

**What is your unfair advantage?**  
Rayford AI's founder and technical advisors have already built and guided a
connected research base in bi-temporal street-view damage assessment,
visual-language disaster perception, CLIP-based multimodal arbitration, and
satellite-to-street disaster imagery. A well-funded competitor can buy data and
hire engineers, but cannot quickly replicate years of domain-specific GeoAI
research, public datasets, model workflows, committee-level technical guidance,
and academic credibility in disaster resilience. The research has also received
industry-facing validation: Mosaic featured the Texas A&M team's Hurricane
Milton damage assessment work in April 2026, highlighting its relevance to
disaster response, insurance risk modeling, and high-quality street-level
geospatial data.

## Section 3 | Market

**Target Market**  
Beachhead: small to mid-size property insurers, claims adjusting firms, and
local governments in hurricane, flood, wildfire, and severe storm regions.
Broader market: national insurers, reinsurers, emergency management agencies,
city resilience offices, infrastructure owners, climate risk platforms, and
property data providers.

**What is your estimated market size?**  
Working bottom-up estimate for validation:

- TAM: $5B to $10B annual global market for catastrophe analytics, claims
  intelligence, property risk data, and disaster recovery decision support.
- SAM: $500M to $1B annual U.S. market for AI-assisted property damage
  assessment and claims triage.
- SOM: $2M to $5M annual revenue within five years from 20 to 50 customers
  paying $25K to $150K per year, plus paid event-based deployments.

These numbers are preliminary and will be tested during AggieX customer
discovery.

**What were your assumptions and methodology in estimating market size?**  
The estimate uses a bottom-up approach. We assume an initial customer base of
insurers, adjusting firms, local governments, and disaster recovery consultants
with urgent property-level damage assessment needs. We estimate annual contract
values from $25K for small pilots to $150K or more for recurring enterprise
deployments. We compare this to the scale of annual disaster losses reported by
NOAA and Swiss Re to confirm that even small efficiency gains in claims triage
and recovery operations can support a large software and data market. These
assumptions need direct validation with customers.

**How many customers have you spoken with?**  
[TODO: enter exact number. If no commercial customer interviews yet, write 0 and
state that AggieX will be used to complete 40 interviews.]

**What key learnings did your customers provide from these interviews?**  
[TODO: replace after interviews. Early hypothesis: customers do not want a black
box that replaces adjusters. They want fast triage, evidence, confidence, and
workflow integration.]

**Describe your competitive moat over time.**  
Rayford AI's moat can compound through proprietary before-and-after property
evidence, labeled disaster imagery, model performance on rare events, customer
workflow integration, and audit history. Over time, each deployment can improve
the property-level resilience graph: pre-event condition, hazard exposure,
post-event change, human review outcome, and recovery status. The company will
also evaluate patent filings around multimodal damage arbitration and auditable
evidence packaging.

## Section 3.1 | Competitors

**Competitor #1: Arturo**  
Approach: AI property analytics from aerial and geospatial data for insurers and
property markets.  
Competitor's Edge: Insurance industry focus, property data workflows, existing
enterprise relationships.  
Your Edge: Rayford AI starts with disaster-specific before-and-after evidence,
street-view damage research, and auditable post-event claims triage.

**Competitor #2: Betterview**  
Approach: Property intelligence for insurance underwriting and risk management.  
Competitor's Edge: Strong insurance positioning and property condition data.  
Your Edge: Rayford AI focuses first on post-disaster damage assessment and
claims triage, then expands into pre-event resilience intelligence.

**Competitor #3: CAPE Analytics**  
Approach: AI-powered property intelligence from imagery for insurers.  
Competitor's Edge: Established property analytics market presence and carrier
relationships.  
Your Edge: Rayford AI uses multimodal disaster reasoning, pre/post event
comparison, and property-level evidence reports designed for disaster response
and recovery.

## Section 4 | Traction & Progress

**Current Traction: Sales & Revenues**  
$0

**Current Traction: Investments/Prizes**  
[TODO: enter exact amount. Include research awards only if allowed by the
application. Otherwise enter $0.]

**Your MVP**  
Current MVP assets exist as research code, datasets, and model workflows for
street-view disaster damage assessment, visual-language damage perception,
multimodal arbitration, and satellite-to-street generation. The commercial MVP
will wrap these into one workflow: select a disaster area, ingest imagery and
property records, generate damage scores and evidence, and export a claims or
inspection triage report. It does not yet have a production web dashboard,
customer login, billing, or enterprise integration. A Mosaic industry article
has already described the underlying Texas A&M research as a rapid hyperlocal
damage assessment tool using pre-event and post-event street-view imagery from
Hurricane Milton.

**Biggest Existential Risk**  
The biggest risk is that the product solves a real technical problem but does
not match a paid insurance or government workflow. We are mitigating this by
making customer discovery the first AggieX milestone, targeting 40 interviews,
testing willingness to pay, and shaping the MVP around triage and evidence
support rather than full automation of claims decisions.

## Section 4.1 | Product Roadmap

**What do you need to build or improve?**  
We need to convert research code into a repeatable Ray Assess workflow. The
workflow must ingest imagery, link it to properties, run damage scoring,
generate confidence and evidence, display results on a map, and export a report
for adjusters or local officials.

**Estimated timeframe**  
10 weeks for the AggieX MVP: May 26 through August 7, 2026.

**How will you measure success?**  
Success means 40 customer interviews, 3 pilot or LOI conversations, 2 validation
case studies, a working web demo, measurable model accuracy on a historical
damage dataset, and at least one customer willing to review a pilot proposal.

## Section 5 | Business Foundation

**Revenue Model and Primary Revenue Streams**  
Rayford AI will sell paid pilots, annual SaaS subscriptions, event-based
deployments, and data or API licenses. Initial pilots may be $2K to $10K.
Annual contracts may range from $25K to $150K depending on customer size,
covered properties, event volume, and workflow integration.

**Prior Funding: Sources, Amounts, Structure**  
[TODO: enter exact funding history. If none, write: No company funding to date.
No SAFE, note, equity, or grant funding has been issued to the company.]

**Use of Milestone-Based Capital**  
Capital removes three bottlenecks: customer discovery travel, product
engineering time, and access to imagery or property data. Funding accelerates
the conversion from research prototype to pilot-ready web demo, supports
customer interviews with insurers and local governments, and helps produce the
first validation case studies.

## Section 5.1 | Business Foundation Uploads

**Business Plan Upload**  
Not complete yet. This repository includes early market, customer, and product
strategy notes.

**Financial Model Upload**  
Not complete yet. To be built during AggieX.

**Pitch Deck Upload**  
Not complete yet. To be built from this application draft and the Rayford AI
website narrative.

## Section 6 | Team Assessment

**What are your team's greatest strengths?**  
The team has deep technical strength in GeoAI, disaster resilience, street-view
imagery, visual-language models, multimodal damage assessment, computer vision,
and the built environment. Yifan has already built and released related research
assets instead of starting from a slide deck. Dr. Lei Zou strengthens the
venture as a scientific and technical advisor, while Dr. Zhengzhong Tu and Dr.
Heng Cai add technical guidance from Yifan's committee. The work is directly
relevant to the first product wedge: property-level disaster damage assessment.

**What are your team's most significant weaknesses?**  
The team is currently stronger in research and prototyping than sales,
insurance procurement, pricing, and enterprise workflow integration. We also
need more direct customer interviews and insurance industry advisors. Because
the founding and advisory team is rooted in university research, we must also
clarify IP, conflict-of-interest, and commercialization structure before pilots
or investment. AggieX is important because it can force the commercial
discipline that the technology now needs.

**What is the biggest failure your team has experienced?**  
[TODO: write a specific story. Recommended angle: a research prototype or broad
GeoAI vision that was technically promising but too broad for a customer to buy.
Explain what changed: narrower customer, clearer workflow, measurable pilot.]

**What is your team's most significant achievement to date?**  
The most relevant achievement is building a connected body of disaster GeoAI
research, code, and datasets around bi-temporal street-view damage assessment,
visual-language disaster perception, and multimodal damage arbitration. This
gives Rayford AI a concrete technical base for the first commercial MVP. In
April 2026, Mosaic published an industry feature on the Texas A&M team's
Hurricane Milton research, naming Yifan Yang and Dr. Lei Zou and explaining why
street-level imagery matters for disaster response and insurance risk modeling.

**Why is this team the right team to win this market right now?**  
This market needs a team that understands both geospatial AI and disaster
resilience. Rayford AI is founded by Yifan Yang, whose research already sits at
that intersection, with technical advisors who can pressure-test the AI,
validation, and built-environment assumptions. The timing is right because
disaster losses are increasing, imagery is more available, and insurers and
governments need faster, auditable property-level decisions.

## Section 7 | Fit & Goals

**Why AggieX?**  
AggieX offers the structure Rayford AI needs now: customer discovery pressure,
mentor feedback, milestone-based capital, and a 10-week execution schedule.
The missing piece is not another research idea. It is a disciplined path from
research prototype to customer-validated pilot.

**What will you accomplish by August 7, 2026?**  
By August 7, 2026, Rayford AI will complete 40 customer interviews, build a Ray
Assess web demo, produce 2 historical disaster validation case studies, define a
paid pilot offer, build an 18 to 24 month financial model, and secure 3 serious
pilot or LOI conversations.

**How did you hear about AggieX?**  
[TODO: referral, McFerrin, social media, class, faculty, or other]

**Anything else we should know?**  
Rayford AI is intentionally starting with a narrow wedge: damage assessment and
claims triage. The larger vision is Ray, a resilience AI assistant that helps
every property understand past, present, and future hazards, prepare before
disaster, and recover faster after impact.

## Section 8 | Commitment

**Attendance Commitment**  
Accepted & Agreed

**Signature**  
[TODO: full legal name of application submitter and primary participant]