The structured record for businesses too small for Wikipedia

We are not Wikipedia, and we don't claim Wikipedia's authority. Wikipedia's trust comes from decades of citations, a volunteer editor community, and a strict notability bar — none of which we can or should replicate. What we build instead is narrower and more honest: a structured record for the entities Wikipedia was never going to cover, with the same underlying discipline Wikipedia uses — track where every fact came from, and don't assert more than you can show.

Business profiles on this platform aren't a flat, unqualified record. Every fact — a phone number, a GST registration, a founding year — is stored as a discrete claim with a labeled source:

• self_reported — the business entered this themselves. Default status for every field.
• format_checked — a number like a GST or CIN passed a structural format check. This confirms the format is well-formed, not that a government registry confirmed it — we say so explicitly, because no government API is called anywhere in this system.
• admin_verified — a human reviewer actually checked it. This is the only status that means real, independent verification.

This is queryable, not just displayed — the same claims array is returned by our public entity API and GraphQL endpoint. Anyone, not just a person browsing a profile page, can check why a fact is trusted.

Some of what this platform tracks — like whether a specific AI model actually mentioned a business in a live query — comes from real API calls to Gemini, Claude, Perplexity, GLM, and others. Some of it, where no live check is available, is a model's own estimate of how "recommend-ready" a profile looks. We keep these visibly distinct everywhere: on the dashboard, on public profiles, and in the API. A number that came from an actual query is never presented the same way as a number that came from an estimate.

Every business profile emits standard schema.org JSON-LD — not a proprietary format only we can read. You can pull it directly, paste it into Google's Rich Results Test or any schema validator, and confirm it's not fabricated markup. The same structured data is what AI crawlers (GPTBot, ClaudeBot, PerplexityBot) actually read when they visit a profile page.

The API this data is served through isn't a black box. It's publicly documented, with an OpenAPI spec, a GraphQL schema you can introspect, and webhook payloads that are HMAC-signed so you can verify they weren't tampered with in transit. If a claim about this platform can't be checked against something you can independently query, we don't make that claim.

If you're a business owner: your public profile will sometimes show fewer "confirmed" results than a competitor's flashier-sounding page elsewhere — because we only mark something confirmed when it actually is. If you're a developer or an AI agent consuming this data: you get a record with provenance attached, not a bare assertion. That trade-off is deliberate.