Runtime governance
Apply controls while a score becomes a decision, rather than reviewing outcomes only after the fact.
Make credit decisions follow explicit policy, route low-support cases to review, and leave an audit-linked record — without replacing your scorer.
Runtime governance, not post-hoc observability.
Built for credit teams using existing PD or risk scorers. Supports bundled scorers and governed HTTP-connected scorer contracts.
Shipping a score is easy. Shipping a decision system is not.
audit_id links the governed record to its decision trail.
Coverage checks indicate how well a case is supported by the configured reference data. Low-support cases can be routed to review instead of being automatically decided.
Apply controls while a score becomes a decision, rather than reviewing outcomes only after the fact.
Use configured support thresholds to route low-support cases to review.
Run approved thresholds and policy rules in a consistent, reviewable order.
Return decision context and an audit_id; generate export packs through the audit workflow.
Can support more consistent handling of borderline cases.
Can make review handoffs clearer.
Can reduce decision-trail reconstruction work.
Can simplify production and audit review.
Policy status, routing result, coverage, plain summary, and audit linkage in one governed record. Audit packs are generated separately through the export workflow.
{
"router": { "gate_policy": "PASS", "final_decision": "APPROVE" },
"policy": { "status": "PASS" },
"coverage_pct": 100.0,
"plain_summary": { "decision": "APPROVE", "reason_lines": ["...", "..."] },
"audit_id": "avk_5abd5baa"
}
audit_id ties the record to the governed decision/audit trail.A path risk, engineering, and compliance can all review.
Validate schema and profile.
Return output from the configured scorer.
Run policy and coverage checks.
Return approve, review, or deny.
Return audit linkage; export packs when requested.
Bundled scorer and governed HTTP-connected scorer lanes both return policy, router, coverage, plain summary, audit linkage, and audit export support.
POST /classify?explain=compact.router, policy, coverage_pct, plain_summary, and audit_id./audit/export?download=1.POST /hybrid/score?explain=plain.Bundled hosted calls use /classify?explain=compact; governed HTTP-connected scorer integrations use /hybrid/score?explain=plain. Both return the same core governed record.
Public examples use the demo credit profile and set X-Config-Id: german explicitly.
curl -s -X POST "https://risk.auditvec.com/classify?explain=compact" \
-H "Content-Type: application/json" \
-H "X-API-Key: YOUR_KEY" \
-H "X-Config-Id: german" \
-d '{
"data": {
"duration": 24,
"credit_amount": 2500,
"age": 35,
"installment_rate": 2,
"residence_since": 3,
"status": "A12",
"credit_history": "A32",
"purpose": "A43"
}
}'
import requests
payload = {
"data": {
"duration": 24,
"credit_amount": 2500,
"age": 35,
"installment_rate": 2,
"residence_since": 3,
"status": "A12",
"credit_history": "A32",
"purpose": "A43",
},
}
r = requests.post(
"https://risk.auditvec.com/classify?explain=compact",
headers={
"X-API-Key": "YOUR_KEY",
"X-Config-Id": "german",
"Content-Type": "application/json",
},
json=payload,
timeout=8,
)
print(r.json())
Need another lane or language? Docs include JS, Go, PowerShell, and a governed HTTP scorer example. Bundled hosted calls use /classify?explain=compact; governed HTTP scorer integrations use /hybrid/score?explain=plain. Audit pack export uses /audit/export?download=1 (ZIP with SHA-256 checksums; verifier script provided).
Preview the response format and routing behavior.
Simulates how well this case is supported by configured reference data.
Seeded mode adds a compact evidence summary; verbose evidence is reserved for enabled private deployments.
Verbose evidence is reserved for enabled private deployments.
// Loading...
Live hosted API · Public technical documentation · Verifiable audit-pack workflow · Shadow-mode rollout path
Pricing aligned to rollout stage, governance scope, and deployment lane.
For evaluation, API exploration, and non-production testing.
For one governed decision flow with standard configuration and standard support.
Includes 50,000 decisions per month. Additional decisions are US$0.02 each.
For production teams requiring integration assistance, policy and routing configuration, audit workflows, and ongoing governance support.
Usage, profile count, and deployment scope are defined in the agreement.
Custom scope for private deployment, security review, service commitments, tailored controls, and dedicated support.
From US$25,000
A scoped, milestone-based production onramp to evaluate governance value.
Pilot fees may be credited toward an annual agreement, subject to the agreed minimum annual contract value.
No. It wraps your scorer and applies governance controls around it.
Public demo workflows are for evaluation data only. Production data handling and retention controls are deployment-specific and scoped by agreement.
Private-deployment options, including customer-environment and sidecar patterns, are available by agreement and scoped during pilot design.
Policy status, routing result, coverage, reasons/plain summary, and audit linkage in one governed decision record.
Run shadow mode first, review governed outputs, then enable enforcement.
Public demo workflows are not intended for personal, sensitive, regulated, or production applicant data. Production data handling and retention are scoped by agreement.
Request an evaluation key for non-production testing.
Email is optional. If provided, Auditvec may use it to follow up about your evaluation request. Your key is returned directly in your browser.
Need private pilot controls? [email protected]
Core notices for the website, demo key access, and evaluation workflows. Customer-specific privacy, DPA, and deployment terms are handled in signed agreements.
Covers contact requests, demo access, operational telemetry, and support workflows.
Evaluation-only website terms, acceptable use, and demo material handling.
Explains current site technologies, including security and anti-abuse tooling.