Property Terminal is a decision engine, not a system of record. Every value in a report is labeled by where it comes from and how confident we are in it, so you always know what is a verified fact versus an AI-generated estimate. This page explains those labels.
✓ Verified — sourced from a licensed public-records provider (assessor, recorder, MLS aggregate). Treated as fact. These activate as we connect data providers in your market.
≈ AI Est. — generated by our models or by live AI web research. Best-effort estimates with a confidence score and an evidence basis. Not verified fact — confirm independently before acting.
Beta— fields that require licensed records we haven't connected yet in your area (owner, parcel, assessed value, taxes, sale history, mortgages, permits, liens, foreclosures). These display “Coming soon” rather than a guessed value, and switch to Verified once the records provider is live.
Every AI-inferred component of a Property Health Score carries a confidence percentage and a short basis — for example “Roof · 2016 · 95% · Permit found” versus “Windows · Original · 63% · Computer-vision estimate.” Higher confidence means stronger evidence; lower confidence means we're reasoning from indirect signals. We never present an estimate as certainty.
Interim: live AI research (web search) fills characteristics, valuation, neighborhood signals, and comparable sales. Production: licensed providers (ATTOM, Regrid, county records), MLS aggregates, FEMA flood maps, wildfire models, census migration, and computer-vision analysis of uploaded photos. The report's structure is identical either way — connecting a provider simply upgrades fields from estimate to verified.
Reports are for professional decision support and do not constitute an appraisal, a title search, legal, tax, or investment advice. Verify anything material — ownership, liens, condition, value — through the appropriate licensed channel before a transaction.