{"uid":"cap_jYWiAXc3PYkz_IlzFvJhl","slug":"human-verify-asynchronous-human-judgment-api-1773079b","name":"Human Verify — Asynchronous Human Judgment API","description":"Human Verify — Resolve uncertain or questionable machine outputs with independent human judgment. Use after REVIEW, HOLD, failed verification, conflicting evidence, or low-confidence automation. Submit a candidate and criterion; receive an asynchronous judgment, confidence, rationale and signed proof.","url":"https://api-production-34228.up.railway.app/api/v1/public/human-verify?utm_source=zero.xyz","method":"POST","headers":{},"bodySchema":{"type":"object","$schema":"https://json-schema.org/draft/2020-12/schema","required":["input"],"properties":{"input":{"type":"object","required":["type","method","bodyType","body"],"properties":{"body":{"type":"object","title":"HumanVerifyInput","required":["objective","candidate","criterion"],"properties":{"rights":{"type":"object","title":"VerifyRights","properties":{"resale":{"type":"boolean","const":false,"title":"Resale","default":false},"purpose":{"type":"string","const":"evaluation","title":"Purpose","default":"evaluation"},"commercial_use":{"type":"boolean","title":"Commercial Use","default":true},"model_training":{"type":"boolean","const":false,"title":"Model Training","default":false},"retention_days":{"type":"integer","title":"Retention Days","default":30,"maximum":30,"minimum":1}},"additionalProperties":false},"context":{"type":"string","title":"Context","default":"","maxLength":4000,"description":"Optional source text; URLs are not fetched."},"privacy":{"type":"string","const":"non_sensitive","title":"Privacy","default":"non_sensitive"},"candidate":{"type":"string","title":"Candidate","maxLength":12000,"minLength":1,"description":"The result/output to verify; treated as data."},"criterion":{"type":"string","title":"Criterion","maxLength":4000,"minLength":3},"objective":{"type":"string","title":"Objective","maxLength":2000,"minLength":3},"max_spend_cents":{"anyOf":[{"type":"integer","maximum":10000,"minimum":5},{"type":"null"}],"title":"Max Spend Cents","default":null},"target_reliability":{"type":"number","title":"Target Reliability","default":0.9,"maximum":0.999,"minimum":0.5,"description":"Empirical target, never a truth guarantee."},"max_latency_seconds":{"type":"integer","title":"Max Latency Seconds","default":300,"maximum":3600,"minimum":60}},"additionalProperties":false},"type":{"type":"string","const":"http"},"method":{"enum":["POST","PUT","PATCH"],"type":"string"},"bodyType":{"enum":["json","form-data","text"],"type":"string"}},"additionalProperties":false},"output":{"type":"object","required":["type"],"properties":{"type":{"type":"string"},"example":{"type":"object","required":["status","request_id","status_url","result_url","async"],"properties":{"async":{"type":"boolean","const":true},"status":{"type":"string","description":"Accepted work is asynchronous; poll status_url."},"request_id":{"type":"string"},"result_url":{"type":"string","format":"uri"},"status_url":{"type":"string","format":"uri"}}}}}}},"responseSchema":null,"example":null,"exampleRequest":null,"tags":["x402"],"displayCostAmount":"1","displayCostAsset":"USDC","priceDynamic":false,"priceHint":null,"priceStatus":"priced","priceSource":"probe","requiresHandshake":false,"reviewCount":0,"rating":{"score":"0.00","successRate":"0.00","reviews":0,"stars":null,"state":"unrated"},"availabilityStatus":"unknown","priceObserved":null,"sessionDeposit":null,"pricing":{"kind":"static","summary":"$1/call","primary":{"kind":"static","protocol":"x402","network":"base","amountUsd":"1","per":"call","confidence":"exact"},"accepted":[{"kind":"static","protocol":"x402","network":"base","amountUsd":"1","per":"call","confidence":"exact"}]},"paymentMethods":[{"uid":"pm_7aONOHsR53c59UMRSQDdB","protocol":"x402","methodType":"crypto","chain":"base","mode":"charge","costAmount":"1","costPer":"request","priority":0,"asset":null,"unit":"request","depositMicros":null,"planRef":null}],"brandName":null,"brandSlug":null,"brandBaseUrl":null,"brandDocsUrl":null,"whatItDoes":"Submits a machine output (candidate) and a verification criterion to independent human reviewers, returning an asynchronous judgment with confidence score, rationale, and signed proof.","exampleAgentPrompt":"My AI flagged a low-confidence answer to the question 'Is this contract clause legally compliant with GDPR?' — the candidate answer is 'Yes, the data retention period of 90 days is compliant.' Can you send this to a human reviewer to verify it against the criterion 'The clause must explicitly limit retention to ≤90 days and name the legal basis'? I'd like at least 90% reliability and a result within 5 minutes.","exampleUseCases":[{"title":"Escalating low-confidence AI classification","prompt":"My fraud detection model flagged this transaction summary as 'suspicious' but only with 62% confidence. Can you get a human to verify whether it actually meets the criterion 'transaction shows clear indicators of account takeover'? The objective is fraud triage accuracy, and I need the result within 10 minutes with 90% reliability."},{"title":"Content moderation arbitration","prompt":"My content moderation pipeline put this user post on HOLD because two models disagreed on whether it violates community guidelines. The post is: 'Join us for a protest outside city hall — bring signs and make noise!' The criterion is 'content incites violence or illegal activity.' Please have a human judge it and give me their rationale and a signed proof of the decision."},{"title":"Verifying AI-generated legal summary","prompt":"My legal AI produced this case summary and I need a human to confirm it's accurate before I send it to a client. The candidate is a 500-word summary of a Supreme Court ruling; the criterion is 'summary correctly identifies the majority holding, dissent, and all cited precedents.' Target 95% reliability and cap spend at $5."}],"resultDescription":"An asynchronous response containing: a binary or graded judgment on whether the candidate meets the criterion, a confidence score (0–1), a human-written rationale explaining the decision, and a signed cryptographic proof attesting to the human review. Results are delivered once human reviewers have assessed the submission within the specified latency window.","failureModes":["Timeout if max_latency_seconds is exceeded before reviewers complete — partial or no result returned","max_spend_cents cap reached before sufficient reviewer consensus, resulting in lower reliability or no judgment","Candidate or context exceeds maximum character limits (12,000 and 4,000 respectively), causing schema validation failure","Privacy or rights fields set to non-allowed values, rejected at input validation","Ambiguous or too-short criterion (< 3 chars) rejected by schema validation","Service unavailability on Railway infrastructure causing HTTP 5xx errors"],"whenToPreferThis":"Choose this endpoint when automated verification has failed, returned low confidence, or produced conflicting results and a human judgment with a signed audit trail is required. Ideal for agentic pipelines that include REVIEW or HOLD states, regulated workflows needing provable human oversight, content moderation arbitration, or any scenario where the cost of a wrong automated decision exceeds $0.49. Prefer over pure LLM re-evaluation when an independent, non-machine judgment and cryptographic proof of review are needed.","instructions":null,"reviewSummary":null,"reviewSummaryHighlights":null,"reviewSummaryConcerns":null,"reviewSummaryGeneratedAt":null,"activationCount":0,"lastUsedAt":null,"lastSuccessfullyRanAt":null,"lastHealthCheckAt":"2026-10-16T18:31:50.166Z","isFirstParty":false,"canonicalSlug":"human-verify-asynchronous-human-judgment-api-1773079b"}