Plain answers, stated exactly as the system works today.
Your PDF is sent over an encrypted connection (TLS) to our server and forwarded to Anthropic's Claude API, which performs the analysis. We do not write your PDF to disk or store it in our database. Only the structured results — the extracted claim rows, totals, and the written summary — are saved to your account so you can revisit, export, or share them.
Analysis is performed by Anthropic's Claude API under their commercial terms, which state that API inputs and outputs are not used to train their models. See Anthropic's commercial terms and privacy documentation for their data-handling specifics.
The sample analysis on the try page uses entirely synthetic data — a fictional insured, fictional claims. No real policyholder or claimant information is involved. You can evaluate the product fully without uploading anything real, and if you prefer, evaluate with a redacted or synthetic loss run.
We use a small set of infrastructure providers — Anthropic (AI analysis), and, in production, Supabase (database) and Vercel (hosting) — and share only what each needs. The full list is in our Privacy Policy.
We are pre-launch and do not yet hold SOC 2 or similar certifications. The analysis is AI-generated and should be verified against the carrier loss run before any underwriting or binding decision — every report says so on its face. If your agency's vendor-due-diligence process needs specifics beyond this page, email hello@losslens.app and we'll answer directly.
See also our Privacy Policy and Terms of Service. This page describes current practices and is not legal advice.