/search, it does not run
website analysis, sanctions screening, or report fetches. It is built to be
called repeatedly as the user types.
How it works
- The query is classified — a name, a BN, or a registration number
- Multiple registries are queried in parallel, each with a short timeout
- Results are deduplicated, merged, and ranked (exact identifier matches first)
- A slow or unavailable source is simply omitted — it never blocks the response
Pre-fill responses typically return in under a second. Debounce input on
the client (~300ms) and cancel in-flight requests as the query changes.
Request parameters
Examples
By name
By business number or registration number
Identifier-shaped queries are auto-detected; the exact match is pinned to the top.What you get back
Each candidate is a canonical profile withlegalName (upper-cased),
operatingNames, registrationNumbers, entityType (normalized to a canonical
set, with the registry’s original wording preserved in sourceEntityType),
status, addresses, an incorporationDate when available, a matchScore
(0–1), and the sources that contributed.
status reports only the current state (e.g. Active) — pre-fill is a fast
identity lookup, not a point-in-time verification, so there is no per-field
timestamp. Use the response-level searchedAt if you need a freshness marker.
Alongside candidates, the response carries candidateWebsites — official
websites harvested for the query and ranked by match confidence, including ones
listed on registry records that didn’t make the candidate list. It is always
present, and empty when nothing was found.
The response also reports mode (how the query was read) and which sources
responded (sourcesSucceeded) or were skipped (sourcesFailed).