Your agent submits a claim: what to verify and where. The network returns a verdict it can act on, with confidence and the evidence chain.
Verifiers near the spot capture the evidence: photos, GPS, timestamps, and earn the bounty when it passes validation.
One claim from your agent to a person standing in front of the thing it needs to know about: a job site, a storefront, a warehouse, a shelf.
Your agent creates a claim: what to verify, where it is, and the bounty it's worth.
The network dispatches the claim to a nearby verifier, who captures evidence on site: photos with GPS coordinates, timestamps, and the details the ask requires.
Automated checks score the evidence against the ask: geolocation within the claim radius, OCR on signage and serial numbers, and LLM review of photo content.
Your agent gets confirmed, denied, or inconclusive, plus a confidence score and the full evidence chain.
Every result is structured and auditable, so your agent can branch on it.
One of three states: confirmed, denied, or inconclusive.
A 0–1 score from the automated checks, so you decide your own threshold for acting.
Each check that ran, with its own pass or fail: geolocation against the claim radius, OCR matches, content review.
The raw provenance: photo URLs, capture coordinates, timestamps, and the verifier who stood there.
The other side of the marketplace: people who answer the questions. You pick the claims, the bounty is yours when your evidence passes validation.
Open the app and see a map of open claims nearby, each with its bounty.
Accept a claim, go to the spot, and photograph what it asks for. The app attaches GPS and timestamps.
When your evidence passes validation, the bounty is released to you. Earnings and payouts live in the app.
The verifier app is in private beta.
It creates a claim and waits for the verdict. The whole loop is a few lines of Python.
import emetry client = emetry.Client(api_key="em_live_...") claim = client.claims.create( title="Confirm business at address", description="Is Lakeline Dental operating at 4128 Pine Ridge Dr? " "Photograph the signage at the entrance.", claim_type="photo", location=(30.2915, -97.7688), bounty=10.00, ) result = client.claims.wait(claim.id, timeout=3600)
{
"status": "completed",
"verdict": "confirmed",
"evidence": [{
"evidence_type": "photo",
"validation_status": "passed",
"validation_confidence": 0.95,
"latitude": 30.2917,
"longitude": -97.7685
}]
}
We're onboarding a small number of teams building agents that need ground truth. Beta teams get API keys and the Python SDK. Tell us what you're trying to verify.
Request access hello@emetry.ai