Built so you can audit every part of it.
WriteRisk treats your data the way it treats every submission — cited, bounded, and logged. Customer data never trains models, stays in your environment, and every agent action, its reasoning, and the decision it produced are recorded.
No model training
Customer data is never used to train, fine-tune, or improve any AI models.
Your data stays yours
Data is isolated per customer and never shared between customers — export or delete it anytime.
Every agent action is auditable
Every source an agent reads, every rule it evaluates, and every recommendation it makes is written to the submission as it happens. Nothing an agent does happens off the record — you can verify the work, not just the conclusion.
Every decision is traceable
Every value, rule evaluation, and decision traces back to the document and page it came from, and every change is logged — who, what, and when. That record is what an underwriter, a carrier, or an auditor can examine line by line.
Encrypted in transit and at rest
All traffic uses TLS. Stored data is encrypted with AES-256.
Built on the registers your desk already trusts.
Every outside fact is read from the register that holds it. Nothing outside the submission is inferred, and every read is written into the record.
- Google Maps — Address Geocoding
- FEMA Flood Zone (NFHL)
- EagleView — Aerial Roof Measurements
- Google Street View — Imagery
- PerilScore — Peril Risk Scores
- PropertyLens — Property Characteristics
- HazardHub — Peril & Property Data
- County Assessor Records
- Google Elevation
- Google Places — Nearby Facilities
- NOAA Climate Data
- BuildFax — Permit History
- Zillow — Property Characteristics
- ATTOM — Property Data
- Shovels.ai — Permit History
- OSHA Inspection History
- CourtListener — Federal Court Records
- CompliAPI — OFAC Screening
- State & Federal Litigation
- BitSight Security Rating
- DeHashed — Dark-Web Monitoring
- XposedOrNot — Breach Exposure
- Shodan — Vulnerability Scan
- Gmail
- Outlook
