A company data API turns a domain into one structured company record — legal or brand name, tagline, description, logo, industry, founding year and headquarters, plus emails, phones, postal addresses and social profiles.
It builds the profile from the company's own website: the homepage and its about and contact pages, read for JSON-LD, page metadata and visible text, with an optional AI extraction pass for the fields that are stated in prose rather than marked up. One domain in, one profile out — which is why the cap is a single company per run and why this is the most expensive unit in the catalogue at $0.03.
$0.03 per delivered profile, up to 1 profiles per run. Nothing delivered means nothing charged.
Give it a domain. It loads the homepage and the about/contact pages, then assembles the profile from three layers, in order of confidence: structured data first (JSON-LD Organization blocks, OpenGraph and standard meta tags), then explicit patterns in the page, then — when use_ai is enabled — a language-model pass over the About text.
That last layer is opt-in for a reason, and the reason is honesty about provenance. Founding year, industry and headquarters are frequently stated in an About page as prose ("founded in Berlin in 2016 by two engineers") and never marked up anywhere a parser can reach. AI extraction reads that sentence. It is also the layer capable of being confidently wrong, so ai_used is returned on every record: you always know whether a field came from markup or from inference. pages_used lists the URLs the profile was built from.
domain is required and is the only identifier — no company name lookup, no fuzzy matching, because a domain is unambiguous and a name is not. use_ai enables the language-model extraction pass for fields stated in prose; leave it off if you want structured-data provenance only. country sets the exit region for sites that vary content geographically.
One record per domain. name is the legal or brand name as the site presents it and tagline its own one-line positioning — often the most useful single field for segmentation, because it is how the company chooses to describe itself rather than how a taxonomy files it. industry, founded and headquarters are the classic firmographics. emails, phones, addresses and socials carry the contact layer. language is the site's primary language, a decent proxy for its home market. ai_used and pages_used are the provenance fields — keep them.
$0.03 per delivered profile ($30 per 1,000). Nothing delivered means nothing charged, and the $2 monthly free credit covers roughly 66 profiles here. Volume tiers take up to 30% off.
$0.03 per profile, $30 per 1,000 — the highest unit price here, because one profile is several page fetches plus an optional model call, delivered as a single assembled record.
Enriching 1,000 target accounts is $30, once. Compared with a firmographics subscription measured in thousands per year, the arithmetic only favours the subscription when you need coverage this cannot give — verified revenue, headcount, technographics, corporate hierarchy. The free $2 monthly allowance covers about 66 profiles.
Two real alternatives, each better than this at something specific.
Official registries — Companies House in the UK, OpenCorporates as an aggregator, national business registers — hold authoritative legal data: registration number, incorporation date, officers, filings. That is the ground truth for compliance and KYC, and nothing scraped from a marketing site substitutes for it. What registries do not have is what the company actually does today: the legal name is often a holding entity, the registered address is often an accountant's office, and the SIC code was chosen at incorporation and never revisited.
Commercial B2B databases hold verified headcount, revenue estimates, technographics and hierarchy across millions of companies. They cost thousands a year, skew toward larger companies in major markets, and their records age between verification passes.
This sits between them: current, self-published positioning for any company with a website, at three cents. For legal facts use a registry; for verified financials use a database; for "what is this company, in its own words, right now" the website is the primary source.
Everyone builds this eventually, and the parts that take longer than planned:
Organization blocks; most have partial ones; many have none.A list of domains becomes a list of companies with industry, size signals, location and contacts — the standard input to routing and scoring.
industry, headquarters and language are enough to assign an inbound signup to the right territory and segment automatically.
Founding year and self-description across a sector map who is doing what, from the primary source rather than a stale taxonomy.
A structured record from a company's own site, with pages_used as the evidence trail — pair it with a registry lookup for the legal facts.
One company per run, by design: this is an assembled profile rather than a list. A domain list means one run each, parallel up to your plan's rate limit — 60 requests/minute on pay-as-you-go, up to 1,200 on the top tier. Sites that are entirely client-side rendered, or that publish nothing about themselves, produce thin profiles; pages_used shows what was available.
When use_ai is on, treat inferred fields as inferred. ai_used exists so that a downstream system can weight them differently, and for anything consequential — compliance, contracts — a registry lookup is the correct check rather than a model's reading of an About page.
Legally: company information published on a company's own website is public and collecting it is generally lawful in most jurisdictions. Corporate data largely sits outside personal-data law, though a named founder or a sole trader's details do not. Marketing use of the collected contacts is governed separately by GDPR, PECR, CAN-SPAM and CASL. Not legal advice.
OpenCorporates offers limited free access to registry data, and several national registers publish free lookups — those give authoritative legal facts rather than current commercial positioning. Here there is a free allowance of $2 a month with no card, about 66 profiles at $0.03 each.
$0.03 per profile — $30 per 1,000 — the highest unit price in the catalogue, because one profile is several page fetches plus an optional model call assembled into a single record. Volume tiers take up to 30% off.
No — the input is a domain, deliberately. A domain is unambiguous; a company name is not, and name matching introduces exactly the class of silent error that makes an enrichment pipeline untrustworthy. If you only have names, resolve them to domains first with the search collector.
It enables a language-model pass over the About text for fields that are stated in prose rather than marked up — founding year, industry, headquarters. Those are frequently written as a sentence and never appear anywhere a parser can reach. The record always returns ai_used so you know whether a field came from markup or from inference.
It is as accurate as what the company publishes about itself, which is usually current and occasionally aspirational. Structured-data fields are high confidence; AI-extracted fields are inference and flagged as such. For legal facts — registration number, officers, incorporation date — use a company registry, which is the authoritative source.
Databases give verified headcount, revenue estimates, technographics and corporate hierarchy for thousands a year, with coverage skewed to larger companies in major markets and records that age between verification passes. This gives current self-published positioning for any company with a website, at three cents. They answer different questions.
No — one domain per run, because the output is an assembled profile rather than a list. A domain list means one run each; they are independent, so they parallelise up to your plan's rate limit rather than queueing.
Skip the marketing. QuantumProxies.io publishes a machine-readable site map, Markdown for every page, a free MCP server, and APIs billed from the same balance as your proxies.
Connect in one command: npx -y quantumproxies-mcp