# TrustThread.ai

> TrustThread.ai is AI provenance infrastructure. It gives every AI agent output a structured evidence chain — a Thread — showing exactly what the output is based on, what was inferred, what was assumed, and where the trail goes dark.

TrustThread.ai is pre-launch. This site is a waitlist / early-access page for builders, operators, and partners. There is no public product, API, or documentation yet.

## Core concept

Every claim inside a Thread is classified into one of four states:

- **Confirmed** — grounded against something real.
- **Inferred** — reached through traceable reasoning.
- **Assumed** — openly assumed, not verified.
- **Opaque** — the agent didn't show its work.

The point in a Thread where solid evidence turns into assumption is called the **Fade Point**.

## Vocabulary

- **Groundline** — the deepest confirmed node in a chain; how far the evidence actually goes before it stops being verified.
- **Fade Point** — the exact node where a chain of solid evidence quietly turns into assumption.
- **Provenance Debt** — the slow accumulation of ungrounded claims an agent racks up over time.
- **Shadow Assumption** — a hidden assumption buried inside a claim that looks confirmed but isn't.
- **Human Validation Gate** — a formal, named human sign-off on a Thread, recorded append-only and designed for cryptographic signing, before a high-stakes output ships.
- **Opacity Ratio** — the share of a Thread's evidence the agent didn't show.
- **Witnessed Ratio** — the share of a Thread recorded by the instrumentation layer rather than asserted by the agent.

## Quality scoring

Thread quality is scored over the output's claims, not raw node counts: depth (how far grounded reasoning actually reaches), confirmed ratio (share of claims grounded in witnessed evidence), opacity ratio (share of claims with no recorded support), and witnessed ratio. Grounding uses weakest-link propagation — a claim is only as strong as its weakest premise. The formula is designed to be open, versioned, and independently recomputable.

## Pages

- [Landing page](https://trustthread.ai/): product overview and early-access signup.
- [Vocabulary](https://trustthread.ai/vocabulary/): canonical definitions of the TrustThread provenance vocabulary — groundline, fade point, provenance debt, shadow assumption, human validation gate, opacity ratio, capture mode, witnessed ratio, grounding, load-bearing node. Each page has a precise definition, a visual specimen, and a live example.
- [Thread TT-0001](https://trustthread.ai/thread/tt-0001/): a real, public, interactive Thread of TrustThread's own project — every claim classified, evidence linked, weak nodes marked openly, with a named human validation gate.
- [Whitepaper](https://trustthread.ai/why/): why the provenance layer is the missing infrastructure of the AI agent economy.
- [Partners](https://trustthread.ai/partners/): the partner program — five founding design partner slots with reserved Thread numbers (TT-0002 through TT-0006) for teams shipping agents with consequential outputs, and a direct path for capital partners.
- [The Missing Layer](https://trustthread.ai/missing-layer/): where provenance sits in the AI stack — the category-by-category comparison with observability, evals, RAG citations, and logging. Observability records what an agent did; evals score whether an output was good; provenance records what an output is based on and who is accountable for using it.
- [The Regulatory Clock](https://trustthread.ai/regulatory-clock/): AI accountability regulation classified by evidentiary status — in force (with primary sources linked), enacted with effective dates set, and in motion — each law mapped to the compliance artifact it demands, with a live countdown to the next deadline.
- [Build Log](https://trustthread.ai/log/): receipts, not roadmaps — a dated record of everything that has shipped, every claim linked to the live artifact, every entry pinned to its commit hash. RSS feed at /log/feed.xml.

## Contact

- Early access: request via the form at https://trustthread.ai/#early-access