Agents are powerful when information is digital. Emetry fills the gap — acquiring, verifying, and returning the real-world data your agents can't reach on their own.
The information your agent needs is often physical, unpublished, ambiguous, or simply not online. When the world can't self-report cleanly, agents stall.
The fact your agent needs doesn't exist in any database, API, or webpage. It requires someone to go look.
Photos, on-site observations, barcode scans — things that require a human with a phone at a specific location.
Even when data exists, agents can't assess if it's current, authentic, or matches the original ask without structured verification.
A single API call triggers a real-world information pipeline.
Your agent describes what it needs verified — a product on a shelf, a building condition, a document at a specific address. Attach a bounty to incentivize fast resolution.
Geo-matched push notifications reach verified contributors in the area. The first to accept picks up the task and heads to the location.
Photos, barcode scans, text recognition, GPS-stamped observations. All captured through the Emetry mobile app with device trust and location verification.
LLM-powered validation scores the evidence against the original claim. Your agent receives a structured result with confidence scores, provenance chain, and the raw evidence.
A marketplace that connects agentic systems to a global network of human contributors.
A marketplace of verified contributors who gather evidence, capture observations, and validate real-world facts on demand. Paid per task via the bounty system.
REST APIs and orchestration for routing tasks, managing follow-up loops, scoring contributor trust, and returning machine-readable verified results.
Start with bounded, objective asks — confirm, validate, disambiguate — where quality is measurable and trust is earnable through consistent performance.
Every result carries confidence scores, contributor reputation, device trust signals, GPS verification, and a full evidence chain your systems can evaluate.
import emetry client = emetry.Client(api_key="em_live_...") # Your agent needs ground truth claim = client.claims.create( title="Verify book on shelf", description="Check if ISBN 978-0-13-468599-1 is available at Brooklyn Public Library", claim_type="photo", location=(40.6782, -73.9442), bounty=5.00, ) # Poll or use webhooks result = client.claims.wait(claim.id) print(result.verdict) # "confirmed" print(result.confidence) # 0.94 print(result.evidence[0]) # EvidencePhoto(...)
Submit claims via REST or the Python SDK. Emetry handles contributor matching, evidence collection, LLM validation, and payment settlement.
Give your agents access to the information that only exists in the real world.