OriginProof is Hive’s market-facing name for Human-Origin Attestation. It turns verified human work, expert review, and dataset provenance into signed receipts a buyer can verify without trusting the platform’s own assertion. Independence is the product: the party that produces the data is not the party that signs the conditions.
Hive doctrine: We Make Everything Provable.
Self-serve today — pay with x402, request an activation key, or run a sample receipt. No meeting, no sales call.
Why independence matters
Buyers of expert data are drowning in a single question: was this actually produced by a credentialed human, under the conditions you say? A seller’s own dashboard cannot answer that — it is the seller vouching for the seller. OriginProof separates the two roles.
“Our humans are real and our process is clean.” The party with the commercial interest in the answer is also the party asserting it. A downstream auditor has nothing to independently re-check.
Hive does not produce the data, employ the expert, or judge the commercial outcome. Hive signs the conditions of production into a portable receipt anyone can verify offline — without calling the platform.
The separation is structural, not cosmetic. The independence that disqualifies a platform from grading its own work is the same independence that makes an OriginProof receipt worth something to the buyer. “Trust our experts” becomes “verify our experts.”
OriginProof binds a signed record of the production conditions into the same provenance graph as the delivered work. Six conditions, one receipt, buyer-verifiable — no change to what the expert actually does.
01
A credentialed contributor cleared against the platform’s own verification, bound to the work.
credential_clearance02
The tools and model assistance permitted for this task, declared up front — not assumed.
tools_declared03
The work happened in one attested session and environment, not stitched from unknown sources.
session_integrity04
The contributor and cohort are bound into the dataset’s provenance graph, not floating metadata.
provenance_binding05
AFiR emits a receipt for each runtime move; AFiR-Stream covers voice and streaming turns — binding the declared conditions to the actual output.
output_binding06
One ML-DSA-65 signed receipt the buyer verifies offline against a pinned key — no call back to the platform.
signer · verify_offlineSee it sign
Each condition is signed independently, then bound under one post-quantum signature. Note the does_not_assert block — the receipt is explicit about what it never claims.
{ "product": "OriginProof", "attests": "provably human-conditioned process", "credential_clearance": "verified", "tools_declared": "disclosed", "session_integrity": "intact", "behavioral_consistency": "consistent", "output_binding": "afir · stream", "does_not_assert": ["definitely_human", "certainly_uncontaminated"], "signer": "hive · independent · ml-dsa-65", "signature": "— awaiting signing —"}
OriginProof attests conditions, never a verdict. It asserts a provably human-conditioned process — not “provably human output.” It makes undisclosed model assistance detectable and costly to conceal through signed declaration / condition mismatches; it does not make cheating impossible or magically detect all AI use.
The same expert network and the same independent proof layer map to adjacent, harder-to-replicate product lines. Each is a strategic option to review with counsel before external launch — not a claim of what exists today.
Line 01
Provably human-conditioned work, certified outward — to journals, publishers, hiring platforms, marketplaces, and review networks facing the “is this AI?” question.
You already produce human work under attested conditions; here you certify those conditions for others to verify.
Line 02
Signed, portable proof of a verified credential that any third party can check independently — without calling back to the platform that issued it.
The credential verification you already perform to onboard experts, turned into a portable artifact.
Line 03
A signed conformance attestation for expert-reviewed regulated work — a protocol, a legal review, an audit deliverable checked against a rubric, provably.
Experts do more than label; this makes their review a signed, regulated-grade artifact.
Line 04
Provably-sourced, expert-assembled research datasets for evidence packages and regulated submissions, where the provenance graph is itself the compliance artifact.
Whole-dataset provenance for any assembled expert body that must survive audit.
Line 05 · the platform option
AI agents purchase provable human expertise on demand: a SmartMorphAgent™ reaches a step that needs a credentialed human, buys that verification in the marketplace, and receives it against a signed OriginProof receipt — settled on-rail with x402.
The same verified expert network becomes the human-in-the-loop layer the agent economy can call into — with provable sign-off, settled per call.
Who this is for
01 · SUPPLY
Ship your existing human work with an independent receipt buyers can verify — turn “trust us” into evidence.
02 · EVAL
Make day-one eval attestation portable: a signed conformance artifact a downstream auditor re-verifies without your harness.
03 · EXPERTS
Turn credential verification you already do into portable, buyer-checkable proof — on 1,600+ contributor relationships.
04 · AGENTS
Give agents a way to buy provable human sign-off on demand, delivered against a signed receipt and settled with x402.
Get started
Every path below is self-serve and real — sign a live receipt, request an activation key, or buy proof scoped to a mission.