Live now. No account, no login, no call

Two things you can check yourself

Pick an attack on a wire transfer and watch execution refuse. The digests are computed in your own browser. Then take the record it hands you and check the second signature against a published key, again in your own browser.

Break a $40,000 wire Check a countersignature

What the demo does not do. It proves the instruction did not change in custody. It does not prove the answer was right or that any model was honest. The second signer is a separate key, a separate clock and a separate log operated by Hive, so it is not an outside party and the service reports that itself. It signs Ed25519, which is not post quantum. The operator side signs ML-DSA-65 under NIST FIPS 204. The second signer's log does not survive a redeploy today, and it says so too. Patent pending.

Private proof-layer workspace · Mercor × Hive · not indexed
Pipeline integrity · expert provenance

Make expert pipelines independently provable.

Mercor connects professionals to AI training projects: experts who evaluate, improve, and guide large language models across finance, law, medicine, consulting, research, and engineering. Every expert session, work product, dataset, and evaluation can carry an independent receipt a buyer verifies without relying on platform logs alone.

AFiR signs the session and output layer. OriginProof adds Human-Origin Attestation, a human-conditioned production receipt. Independence is the product: platform logs are useful, but independent receipts travel outside the platform. This is pipeline integrity and provable expert provenance, signed at the session, output, and dataset layers.

Self-serve today: pay with x402, request an activation key, or run a sample receipt. No meeting required. This page is meant to answer the question on its own.

Why independence matters

A platform log is what the platform recorded. A receipt is what a buyer can verify.

Both have their place. The difference is who has to be trusted for the record to mean something.

Platform log

“Our system recorded it.”Useful operational truth, but the buyer has to trust the platform that produced it. The evidence and the interested party are the same entity.

Independent receipt

Evidence that verifies outside the platform.A signed receipt a buyer checks against a public key, with no need to trust the marketplace’s own assertion. The record travels with the deliverable.

Hive does not source the expert, run the marketplace, or assess the commercial outcome. Hive signs attested conditions and outputs, so the party that produces the work is not the party that signs the proof. That separation is exactly what makes a receipt worth more than a claim.
Prove now · the pipeline Mercor already runs

These map onto expert work that already happens. There's nothing new to build first. Add a signature layer to the pipeline you run today, and the output becomes buyer-verifiable.

01

Expert eligibility receipt

Attest that the person who did the work cleared the credential and eligibility bar declared for the task, without exposing the underlying identity documents.

credential_clearance
02

Session integrity receipt

Bind the working session (start, continuity, and completion) so a buyer can see the deliverable came from a coherent, uninterrupted expert session.

session_integrity
03

Declared tool / model-use scope

Record which tools and models were declared in scope for the task, making undisclosed assistance detectable and costly to conceal. Not impossible, but detectable.

tools_declared
04

Output binding: AFiR / AFiR-Stream

AFiR emits a receipt for each runtime move. AFiR-Stream covers voice and streaming turns, binding the declared conditions to the actual output the buyer receives.

output_binding
05

Dataset / work-product provenance

Carry lineage through an assembled dataset or deliverable, showing which expert conditions produced which rows, so provenance survives packaging and resale.

dataset_lineage
06

Buyer-verifiable receipt package

Bundle the above into a signed package a buyer verifies against a public key, independent of the marketplace. Proof that travels with the work.

signer · ml-dsa-65
Build next · turn provenance into product surface

Once the receipts exist, provenance stops being an internal control and becomes something you can sell. These are strategic options to review with counsel, not claims that these ship today.

A

OriginProof / Human-Origin Attestation

Certified human-conditioned work as a named product line: a provably human-conditioned process a buyer can price a premium against.

product: OriginProof
B

Imprimatur / portable credential attestation

A portable, signed attestation of an expert’s cleared credentials that follows the work across engagements. Reusable, revocable, buyer-checkable.

attestation: portable
C

Mission Coherence / signed review conformance

Prove that an expert review conformed to the declared rubric or standard: a signed conformance receipt, not just a score in a dashboard.

review_conformance
D

Assembly Receipt / dataset provenance graph

A provenance graph for an assembled dataset, with every contributing expert condition linked and signed, so a research or model buyer can audit lineage end to end.

provenance_graph
E

SmartMorphAgent™ + x402

Agents purchasing verified human expertise on demand: an agent settles with x402 and receives a signed receipt binding the human-conditioned work it just bought.

agent_settlement: x402

See it sign

A sample receipt for a Mercor-style expert task

Click Sign the receipt. Each condition signs in turn and the live receipt fills field by field. Nothing here asserts a route, only the attested conditions of production.

  • expert_eligibilityCredential + eligibility bar cleared for the task
  • session_integrityCoherent, continuous expert session
  • tools_declaredDeclared tool / model-use scope recorded
  • output_bindingAFiR binds conditions to the delivered output
  • dataset_lineageWork product linked into dataset provenance
receipt.json unsigned
"product": "OriginProof",
"expert_eligibility": "credential_cleared",
"session_integrity": "continuous",
"tools_declared": "in_scope",
"output_binding": "afir / afir-stream",
"dataset_lineage": "linked",
"signer": "hive · ml-dsa-65",
"does_not_assert": ["definitely_human", "certainly_uncontaminated"]
Boundary, stated plainly. OriginProof attests conditions, never a route. It asserts a provably human-conditioned process, explicitly not “provably human output.” The receipt does not assert definitely_human or certainly_uncontaminated. It makes undisclosed model assistance detectable and costly to conceal, not impossible.

Start signing

Pay, activate, or see it, without a meeting.

Every path is self-serve and real. Settle with x402, request an activation key, or open OriginProof and run a sample receipt.

Sources & grounding

Public, organization-level grounding for the framing above. No individuals named; claims are kept to what these public materials state.

  • Mercor: AI Training Roles (Experts): professionals apply expertise to evaluate, improve, and guide large language models. mercor.com/experts
  • Mercor Research: Frontier AI Training Data & Human Evaluation: expert-produced training data and human evaluation. mercor.com/research
  • Mercor: What is AI Training Data: expert-labeled data across specialized domains. mercor.com/resources
  • TechCrunch: Mercor coverage: described as an AI talent platform and AI training startup. techcrunch.com/tag/mercor
new in the canon · run the pipeline sequence

Give the evaluation handoff a checkable order

This addition gives an expert work pipeline a record of the evaluation handoff and its order. The receipt is ready in the codebase but awaits deployment, and the examples beneath report that exact state.

eval.administration

Record the order of an expert evaluation handoff

When an expert task moves from selection to testing, the sequence can matter as much as the final work product. This receipt links one evaluation attestation to a named administration arrangement. It locks a test set fingerprint before the candidate or contributor learns its contents and before the scheduled evaluation window opens. The service derives an independence class from the stated relationship and assigned duties rather than taking the classification from the sender. It preserves the claimed administrator with the stated relationship and duty separation as unverified assertions.

What it does not do. It returns neither a score nor an outcome and cannot establish whether the assessment was appropriate or representative, whether participants avoided every other path to its items, whether its method was adequate, or whether the person administering it was competent. It neither has the status of a certification or accreditation nor is it an audit opinion; standards bodies do not currently recognize it, nor do regulators or insurers.

POST /verify/eval-administration · case pass, a clean record

Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.

POST /verify/eval-administration · case fail, a forged record

Nothing has run yet. Click Run it and the answer below comes back from the verifier, not from this page.

Every run above posts a verified request body from this domain to the open verify route and prints what came back. The example receipts are signed with published example keys, so verify reports key_trust example_registry. That is on purpose. Nothing on this page is a production issuance, a customer record, or an endorsement. Patent Pending.