OpenCorporates

OpenCorporates is the largest open database of companies in the world, and its distinguishing feature is provenance: every record traces back to an official government company register rather than a scraped or crowd-sourced guess. A typical entry carries the company name, registration number, jurisdiction, incorporation date, current status, registered address, and, where the source register publishes them, the names of officers and directors. Coverage runs to hundreds of millions of companies across more than 100 jurisdictions.

Individual company pages are free to browse and search on the website, which is the fastest way to see how much detail a given jurisdiction exposes. Anything at scale goes through the API or a bulk data export, and both are commercially licensed. The API returns JSON, which is the sensible route for RAG: resolve a name or number to a canonical record, pull the linked officers, and store the structured fields next to a short generated summary sentence so a retriever has readable text to match on.

Its natural home is the backbone of a corporate knowledge graph: connecting companies to their directors, tracing ownership and control chains, or grounding a due-diligence, know-your-customer, or investigative-research assistant that has to answer who runs a company and where it is registered.

Watch the unevenness. OpenCorporates can only surface what each register discloses, so depth swings from rich officer and filing histories in some jurisdictions to little more than a name and number in others, and freshness follows each register's own update cadence. The licence matters too: this file marks commercial use as restricted, so while browsing is free, any bulk or API reuse needs a paid licence. Budget for it before designing a product around the data.

It dovetails with OpenSanctions, which links its entities back here, and with GLEIF's legal-entity identifiers when you need a shared company ID across datasets.