The TrustThread project, pulled apart node by node
Most projects ask you to trust the confidence. This one shows you the chain. Click any node — the green ones link to their sources, and the yellow, red, and gray ones tell you exactly why they aren't green. A project with only green nodes would be the first sign of a liar.
TT-0001 IS HAND-BUILT AND HUMAN-VERIFIED — IT PREDATES CAPTURE MODES. IN THE PRODUCT, CONFIRMED REQUIRES EVIDENCE WITNESSED BY THE INSTRUMENTATION LAYER, NOT BY US.
Node detail
Human Validation Gate
"This is our own project, so I reviewed every node instead of rubber-stamping the green ones. The market figures are aggregations, not audited numbers — treat them that way. The chain is honest about where it fades. That is the point of the product."
On the record. Append-only. Designed for cryptographic signing. If this decision is ever reversed, a new gate will supersede it — this one never disappears.
Why our own project has red and gray in it
Because that's the product. Two of the market numbers above are aggregations we couldn't trace to one canonical source — so they're yellow, not green. One claim from our own materials is still missing its primary source — so it's red, and it stays red until we find one. One node is founder conviction formed across unrecorded conversations — we can't show what we didn't capture, so it's gray instead of being dressed up as analysis.
Every AI claim you've ever seen — from companies and from AI systems themselves — is presented with the same confidence whether it's strong or weak. A Thread makes that impossible. The fade point in this graph is exactly where our evidence gives way, marked in public, on our own project.
This Thread was built by hand. The product builds them automatically, from live agent runs, at machine scale. If you want the version of the future where every consequential AI output carries one of these: