From a day of filings to a record you can act on.

The fields, and where they come from.

A filing on its own is a caption and an index number. These are the fields we add around it, each one tagged with the source it came from so you can weigh it yourself.

The case. Index number, court, case type, filing date, demand amount where the complaint states one, and a direct link to the docket on the court’s own system. Parsed from the summons and complaint, not retyped.

The parties. Plaintiffs and defendants as named in the filing, with service addresses where the documents disclose them. Business defendants are resolved against the New York Department of State corporate registry and NYC consumer-affairs licence records, then run through the Census geocoder and stored in standardised form. Verified and unverified are labelled differently, and registered-agent addresses are marked as such rather than sold as mailable.

Counsel. Attorneys of record matched against the New York attorney-registration roll: bar number, admission year, registration status, office of record. Delinquent or due-to-re-register status is carried on the record.

The plaintiff’s history. Federal docket counts from the RECAP archive, plus how often that plaintiff and their usual counsel appear across the corpus. A filer with thousands of federal matters is running a template; that is worth knowing before you read the complaint.

What we do not supply. No predictions, no scores estimating how a case will end, no characterisation of any party. The record is the product, and it is worth more when it is exactly what the filing says and nothing more.

You receive records, not documents. Original filings stay on our own machines. What reaches you is the structured record — so an exhibit attached to a complaint, and whatever account numbers or personal details it happens to contain, never travels with it.

Anyone who asked to be left out is already gone. When a person named in a filing asks us to remove them, we delete what we gathered and their case stops appearing in anything we supply — enforced by the software, not by somebody remembering. You are never handed a record that someone has objected to being in.

Not scraped, on purpose.

The courts block automated access, and we do not try to get around it. No bots, no CAPTCHA solving, no scripted collection running behind somebody’s login. An operator sits down and works the court’s own search the way any subscriber does.

That is slower, and it is the point. It means the collection method survives a hard question about where a defendant’s address came from, and it means the source stays available - a data supplier who gets an access route shut off has no product at all.

The automation sits after acquisition, not during it. Corporate registries, licence records, the attorney roll, the federal docket archive and the geocoder are all official open data, queried through published interfaces, cached, and recorded per field. Anything the parser could not read confidently is held for human review rather than shipped with a guess attached.