This page is generated straight from the registry the engine runs on, so it always shows exactly what KOANO queries and can never drift into naming a source we do not use. Most of it is live public data. Some figures are honest stand-ins for licensed data we have not funded yet, and they say so wherever they appear.
Five
Provenance states
Sourced
Every figure, every verdict
Daily
Archive of the public record
KOANO calls authoritative public data at the moment you run an analysis: municipal building records, parcel and zoning data, recorded sales, federal hazard and climate data, environmental contamination records, crime statistics, mortgage and lending activity, employment and migration. National sources run anywhere in the US. The NYC-tagged municipal layers are the deep local record that lets a New York verdict roll up fully live.
Municipal building records
Parcel, zoning and ownership
Recorded sales
Environmental contamination
Federal hazard and climate
Crime statistics
Mortgage and lending activity
Employment and migration
Prices, demographics and housing policy
KOANO runs live everywhere it can. Where a source is licensed commercial data we have not funded, it uses a clearly labeled representative stand-in and names the license that turns it live. These never feed a verdict. They support documents only, and they are never shown as real.
Development documents
Becomes live with a CoStar Market Analytics or HouseCanary license. A one-line change in the provider registry.
Portfolio documents
Becomes live with a CoStar/LoopNet or MSCI Real Capital Analytics license. A one-line change in the provider registry.
A figure is only as trustworthy as where it came from, so KOANO tags each one and keeps the states distinct. A verdict's overall label equals the weakest input it used, so a computed score can never hide a stand-in inside a tidy average.
Fetched in real time from an authoritative public source KOANO queried when you ran the analysis. Full trust.
A third-party partner feed. It carries that partner's trust profile and is attributed to them by name, never shown as something KOANO verified itself.
A labeled placeholder for a licensed source KOANO has not funded. You can inspect it, and it is never shown as real.
A source KOANO covers and attempted, that failed on this run. The figure is shown as missing, never guessed, and it usually clears on a retry.
A market or layer KOANO does not cover yet, such as the municipal record outside NYC. Nothing was queried, and the gap is named honestly.
When you run an analysis, KOANO queries its live sources right then. There is no cached market report sitting behind the answer. Each figure carries the source it came from and the time it was fetched.
Every signal is normalized to a common shape: location (lat/lng and census tract), timestamp, provenance state, and source attribution. Nothing enters the reasoning without a label.
Records are stored in Supabase with Row Level Security enforced, scoping every record to the account that owns it. Dedicated per-tenant isolation is on the enterprise roadmap, not live today.
Separately, a daily job snapshots the free public record, such as permits, violations, zoning, and ownership, into a weekly time series the source datasets themselves do not keep. Only live data is ever archived.
The assembled signals go to the five specialist agents at once. The synthesis agent arbitrates and issues one verdict, whose overall provenance equals the weakest input it used.
Every KOANO verdict cites its sources and labels every figure. Sign up, run three analyses free, and check the work against an address you already understand.