Model-backed rail engine
Model-backed Compliance EvidenceOps analysis.
Agents submit structured compliance packet context. Wever Labs calls Model-backed for evidence summary, missing evidence, risk notes, readiness rationale, and next action, then binds the analysis to a Wever Labs proof object.
Model-backed analysis enabled
Return package
Model-backed thinks through the packet. Wever Labs proves the run happened.
API
/api/evidence-analysis
/api/evidence-analysis
Rail
Compliance EvidenceOps
Compliance EvidenceOps
Returns
analysis + receipt passport
analysis + receipt passport
Next
Retry or Analysis Proof Relay
Retry or Analysis Proof Relay
One-click live demo
Packet in. Evidence analysis out. Receipt verified.
Click once to POST a vendor security questionnaire packet to the live endpoint. The response shows the model state, missing evidence, next action, proof hash, and receipt verification state while private provider credentials stay outside the response.
Waiting for sample run.
Run the sample to see the handoff summary.
View sample payload
{
"rail_key": "compliance_evidenceops",
"packet_type": "vendor_security_questionnaire",
"requested_outcome": "readiness_review",
"organization_name": "sample_organization",
"control_framework": "vendor_security",
"required_evidence": ["soc2_report_or_bridge_letter", "insurance_certificate", "security_policy", "subprocessor_list", "data_handling_summary"],
"available_evidence": ["soc2_report_or_bridge_letter", "insurance_certificate", "security_policy"]
}What Model-backed returns
Analysis becomes a proof-backed object.
The analysis engine is intentionally narrow. It accepts structured business compliance packet context and returns operational fields that downstream agents can verify, retry, or relay.
Analysis
Evidence summary
What the submitted packet contains and how it maps to the requested outcome.
Gap
Missing evidence
Required packet items absent from the submitted context.
Risk
Risk notes
Concise operational risks that affect handoff readiness.
Score
Readiness rationale
Why the readiness score is high, partial, or not yet ready.
Action
Recommended next action
Route to Retry Manager when evidence is missing, or continue to Receipt Passport and Analysis Proof Relay.
Proof
Wever proof object
analysis_hash, proof_hash, receipt_hash, receipt_signature, and verification state.
POST shape
Sample agent payload.
Agents can use this as the first model-backed Compliance EvidenceOps test.
{
"packet_type": "vendor_security_questionnaire",
"requested_outcome": "readiness_review",
"available_evidence": [
"soc2_report_or_bridge_letter",
"insurance_certificate",
"security_policy"
],
"required_evidence": [
"soc2_report_or_bridge_letter",
"insurance_certificate",
"security_policy",
"subprocessor_list",
"data_handling_summary"
]
}Boundary and guardrails
One engine inside one rail.
Model-backed does the evidence analysis. Wever Labs still handles the rail contract, receipt passport, proof hash, retry path, ledger compatibility, and downstream handoff. GET requests return sample objects without spending model provider credits. Live model calls happen only on POST and are bounded by body-size and daily rate guardrails.