Every fact on data.techspike.tv links to the public record that asserts it. We do not publish claims we cannot attribute. That lets you verify any result, and lets an agent built on this data cite its sources rather than invent them. We resolve records from dozens of public sources into one view of a business; the rules below govern how we attribute, qualify, and retain what we show.
Every fact is cited
When we report that a company is sanctioned, owned by another, or recalled a product, we name the source that asserted it: the original publisher, the dataset, and the date we recorded it. We assert nothing on our own authority.
Open any source from the catalog to see its official link and its live activity.
Two dates: when it was true, and when we saw it
We keep two timestamps apart, because they answer different questions:
- When the source said it was true— the date the publisher attaches to the record (when a sanction took effect, when a filing was dated). Sources don't always provide this.
- When we recorded it — the moment our system pulled it in. We always have this.
Keeping them separate lets you tell a current designation from a stale one, and see how quickly we picked up a change.
How reliable each source is
Sources differ in authority, so each carries a short note on its credibility: who created it, its type (official regulator, public registry, investigative leak, or aggregator), how often it updates, and its known limits.
A name appearing in the ICIJ offshore-leaks data, for instance, is a lead, not a confirmation: those records often carry no identifiers or address, so we surface them as something to investigate, never as a confirmed identity.
Where we use automated judgment
Two steps in our pipeline make automated judgments, and we disclose both. Record matching — deciding that several records describe the same entity — is algorithmic: it relies on shared identifiers and corroborating details, and every resolved entity shows how it was assembled and at what confidence. The chat assistant is a language model; its answers can be incomplete or vary between runs. In both cases the cited sources are the authority. Verify against them.
We retain each source's full history
Every addition, change, or removal from a source is appended to a permanent log rather than overwritten. After a record disappears upstream, we can still show what it said and when, and reconstruct an entity as of any past date. Each source page links to the upstream original.
The standard we follow
This is not ad-hoc. We apply the U.S. intelligence community's published standard for citing open and public information (ICS 206-01) — the discipline analysts use so that any claim can be independently verified. Attribution is therefore consistent across sources, and a conclusion drawn from this data is one you can defend.
We always show the most detail we can; the only thing we hold back is anything that would expose a credential or access token.