[ blog · use-case ]8 min read

Entity Resolution With LEI: A GLEIF Lookup API for Counterparty KYC and Ownership Trees

Sarah Choy·Published September 1, 2026·8 min read

"Apple Inc.", "Apple, Inc." and "APPLE INC" are three strings and one company. The LEI is a 20-character code that says so unambiguously, in every jurisdiction, and the data is public domain.

TL;DR

• •An LEI is a 20-character ISO 17442 code identifying one legal entity worldwide. Unlike a company name it is unambiguous; unlike a national register number it does not collide across jurisdictions.
• •GLEIF publishes the entire register under CC0 — public domain, no attribution required, no restriction on downstream use. That is rare for entity data and it is the reason this endpoint can exist.
• •Look up by LEI code for an exact record, or search by legal name when you only have a messy CRM string. Name search is relevance-ranked, not exact, so total_matches can be large.
• •registered_as carries the entity's number in its national company register — the bridge from a global identifier to local statutory filings.
• •Level 2 data gives direct parent, ultimate parent, and subsidiary count, each parent identified by its own LEI so you can walk the ownership tree upward or downward.

The join key problem

Every company that has ever tried to reconcile two systems knows this one. Your CRM says Apple Inc., the invoice says Apple, Inc., the sanctions screen says APPLE INC, and the contract says Apple Operations Europe, which is a different legal entity entirely. Fuzzy matching gets you most of the way and then produces a false positive that matters.

The Legal Entity Identifier exists to end this. It is a 20-character code, defined by ISO 17442, that identifies exactly one legal entity anywhere in the world. Anyone trading on regulated financial markets is required to have one, which means coverage of serious counterparties is high.

The part that makes it usable for the rest of us: GLEIF publishes the entire register under CC0 — public domain, no attribution, no restriction on downstream use. That is close to unheard of for entity data.

Two ways in

You have the code

curl "https://www.apipick.com/api/lei-lookup?lei=HWUPKR0MPOU8FGXBT394" \
  -H "x-api-key: $APIPICK_KEY"

{
  "mode": "lei",
  "found": true,
  "record": {
    "lei": "HWUPKR0MPOU8FGXBT394",
    "legal_name": "Apple Inc.",
    "other_names": ["Apple Computer, Inc."],
    "jurisdiction": "US-CA",
    "legal_form": "H1UM",
    "entity_status": "ACTIVE",
    "registered_as": "806592",
    "bic": ["APLEUS66XXX"],
    "registration_status": "ISSUED",
    "next_renewal_date": "2027-03-08T17:27:20Z",
    "headquarters_address": { "city": "Cupertino", "region": "US-CA", ... }
  }
}

Two fields deserve attention. other_names carries previous legal names, which is how you match a contract signed under a name the company no longer uses. registered_as is the entity's number in its national company register — the bridge from a global identifier to local statutory filings.

You only have a messy string

Pass name instead, optionally narrowed by country.

GET /api/lei-lookup?name=Tesla,%20Inc.&country=US&limit=3

{
  "mode": "search",
  "total_matches": 43128,
  "returned": 3,
  "records": [
    { "lei": "54930043XZGB27CTOV49", "legal_name": "TESLA, INC.", ... },
    { "lei": "549300LFXY1VG8DI7B68", "legal_name": "Tesla General Insurance, Inc.", ... },
    ...
  ]
}

Walking the ownership tree

Add include_relationships=true and you get GLEIF Level 2 data: direct parent, ultimate parent, and a count of direct subsidiaries. Because each parent comes back with its own LEI, walking upward is just another call.

import httpx, os
H = {"x-api-key": os.environ["APIPICK_KEY"]}
URL = "https://www.apipick.com/api/lei-lookup"

def climb(lei, depth=5):
    """Walk from an entity up to its ultimate parent."""
    chain = []
    for _ in range(depth):
        r = httpx.get(URL, params={"lei": lei, "include_relationships": "true"},
                      headers=H).json()
        if not r["found"]:
            break
        chain.append(r["record"]["legal_name"])
        parent = r.get("relationships", {}).get("direct_parent")
        if not parent:
            break
        lei = parent["lei"]
    return chain

# ['APPLE OPERATIONS INDIA PRIVATE LIMITED', 'Apple Inc.']

This is the question that matters before you extend credit or sign: not "who am I contracting with" but "who ultimately controls the entity I am contracting with." A subsidiary with a thin balance sheet and a parent with a strong one is a different deal from the same subsidiary standing alone.

Where it fits in an onboarding stack

These three answer different questions and are worth combining rather than choosing between:

• LEI — who is this entity, officially, and who owns it. Global, CC0.
• EU VAT — is this business registered for cross-border trade in the EU. Authoritative, but EU-only and says nothing about identity beyond a name.
• Company Facts — SEC filings and financials for US public companies.

A reasonable onboarding flow resolves the LEI first to fix the identity, then validates the VAT number for the tax treatment, then pulls financials if the counterparty is public.

Signals people miss

registration_status is not decoration. An LEI must be renewed annually, and a LAPSED status usually means the entity stopped trading in regulated markets or stopped paying attention — either of which is worth knowing before a large contract. entity_status: INACTIVE is stronger still: the entity has been dissolved or absorbed, and successorEntity often names what it became.

Build vs. call

GLEIF's own API is free and the data is identical — this is the same register either way. What you take on directly is JSON:API nesting, a 404 that arrives as an HTML error page rather than JSON, and separate endpoints for each relationship type.

Source and licensing

Records come from GLEIF, the Global Legal Entity Identifier Foundation, which maintains the LEI register on behalf of the Regulatory Oversight Committee. Level 1 reference data and Level 2 ownership data are both published under CC0. Three credits a call, charged only on success; a free key comes with 100 credits and no card.

Written by

Sarah Choy

CEO, API Pick

Sarah Choy is the CEO of API Pick. She writes about building production-ready APIs for AI agents and LLM workflows.