Log in

Evidence before conclusions

Methodology

LotContext resolves one NYC property identity, queries bounded public-record sources, preserves each source's coverage state, and links deterministic findings back to the records that support them. It does not use generative AI to decide what a property “means.”

1. Resolve before querying

A submitted address is resolved with NYC GeoSearch. LotContext records the normalized address, borough, BBL, BIN, block, lot, and coordinates that the service returns. Downstream queries use those identifiers; raw address text is not inserted into Socrata filters.

A BBL identifies a tax lot and a BIN identifies a building. They are not silently merged. Condominium lots, multi-building parcels, incomplete identifiers, and conflicting mappings remain visible limitations.

2. Keep parcel evidence separate from context

Records matched by a supported parcel or building identifier are shown as property evidence. Radius, ZIP, precinct, and other administrative-area records are labeled as neighborhood context. Nearby activity is never represented as an event at the searched parcel.

3. Preserve source and retrieval state

Each source is fetched independently with bounded timeouts and retries. The report stores the retrieval time, source identity, source record identifier, match method, raw public payload, normalized fields, and whether the source completed, returned no records, timed out, was unavailable, or was otherwise partial.

A source failure is never converted into “no records.” An empty result never proves compliance, safety, title status, legal occupancy, or legality.

4. Produce explainable findings

The findings engine applies reviewed rules to returned records—for example, flagging an open enforcement record or repeated recent complaints. A finding includes evidence references, confidence and severity labels, and diligence questions. It is research support, not a legal, engineering, title, appraisal, lending, insurance, or compliance conclusion.

5. Cache without hiding age or authorization

Public-source results may be reused for a bounded source-specific freshness window. Retrieval timestamps remain visible, and authorization is checked again whenever a private workspace or frozen report snapshot is opened. Opaque share tokens—not report IDs—control public sharing.