KYC confidence thresholds with Jev
Jumio-class vendors, sanctions lists, and liveness stay the system of record. Jev can ask whether the leftover narrative looks complete or internally inconsistent. It does not verify a passport and it is not a licensed KYC program.
This unofficial page is the confidence thresholds slice of the KYC residual-text decisions pack. Intent: apply the Jev (TypeSafe System One) decision model to KYC residual-text decisions confidence thresholds. Primary search language: KYC Jev confidence thresholds. Confirm patterns on docs.typesafe.ai. This site does not sell, issue, or proxy TypeSafe keys. Use a credential you already have from the console or a documented gateway.
Independent angle (cover ≠ clone): Vendor attests identity; Jev only labels residual application text after structured checks. Fail closed — not a clone of KYC-vendor IA or an “AI KYC” product page. Treat floors as production gates and price the false-reject cost — official 0.5/0.9 sketches are illustrations.
KYC use-case context
Thresholds turn KYC residual-text decisions answers into act / review / abstain. They are product policy, not a hyperparameter TypeSafe ships. Official 0.5 / 0.9 sketches are illustrations. This slice also carries the false-reject discussion: over-gating KYC residual-text decisions hides calibration.
Hub: Use cases. Compare, when the other tool is the real job: KYC vendors.
Confidence Thresholds inputs
You need (1) pinned answers on a frozen contract and (2) labels for hold / proceed / escalate gold from KYC analysts, plus contradiction gold. State shape:
{
"application": { "id": "APP-19", "narrative": "I live in Lisbon but the business is registered in Delaware for tax reasons." },
"vendor": { "id_result": "pass", "sanctions_hit": false, "country_claimed": "PT" },
"policy": { "consistency": "Flag country or occupancy claims that contradict vendor country_claimed." }
}
Decision signals and actions
| Axis | Where it lives | KYC use |
|---|---|---|
choice / score / noul |
answer payload | What to do with the application narrative + vendor result summary |
confidence |
Choice & Score only | Whether to trust the argmax |
| Distance from 0.5 | Noul | Whether narrative_consistent is decided |
FLOORS = {
"park_for_ops": 0.55, # illustrations — replace
"auto_approve_account": 0.95,
}
NOUL_TAU = 0.80 # for narrative_consistent
def allow(ans, action):
return ans.confidence >= FLOORS[action]
Do not treat a Noul of 0.5 as a “medium” KYC residual-text decisions score — it means yes and no are equally likely. Conjunctions stay in your code.
Guardrails and escalation
TypeSafe’s confidence-gated examples use a lower bar for recoverable reads than for irreversible actions. Those numbers are illustrations. For KYC residual-text decisions, treat auto_approve_account as the high bar (opening an account or attesting identity). Tune on labels — see offline evaluation.
Band around 0.5 on narrative_consistent always reviews. Do not copy 0.80 onto Choice confidence.
Evaluation and rollout notes
- False-proceed on planted contradictory narratives (must stay below your SLA)
- Analyst override rate
- Calls skipped because vendor already failed — those are successes of the fail-closed path
Fit loop: pin jev-1.13.0 → replay → plot error vs confidence → pick floors where auto-act error ≤ your SLA. Pin jev-1.13.0 (the versioned id) after you fit thresholds. jev-latest and the marketing line jev-1.13 can move. Log the response model. TypeSafe’s published list price for jev-1.13 is $0.042 per million input tokens (vendor claim — confirm on the models page); output tokens are free on that same page. Unused distractors still bill as input.
Official Python and JavaScript SDKs read TYPESAFE_API_KEY and retry documented 429/529. This site does not sell, issue, or proxy TypeSafe keys. Use a credential you already have from the console or a documented gateway.
Pack map
| Slice | Page |
|---|---|
| Graph and primitives | decision workflow |
What may enter state |
input contracts |
| What to gather first | evidence collection |
| Atomic rules | policy checks |
| Act / review / abstain | you are here |
| Reviewer payload | human handoff |
| What to persist | audit trail |
| How it breaks | failure modes |
| Labeled replay | evaluation |
| Shadow → canary | production rollout |
FAQ
Should auto_approve_account use 0.9 everywhere? No. Over-gating hides calibration and dumps the queue on humans. Fit per action.
Can I reuse a Noul τ as Choice confidence? No. Jaggedness: they are not interchangeable. See confidence.
Where is the rest of the KYC pack? Start with KYC decision workflow and KYC human handoff. Cluster hub: Use cases.
Can Jev replace Jumio / Onfido / our list provider? No. Keep the vendor of record. This page is leftover text triage. See Jev vs KYC vendors.
Is this legal advice or a KYC certification? No. Unofficial field notes. Counsel and your program own attestation.
What this page does not claim
- Not a substitute for KYC/AML programs.
- No pass-rate or false-accept claims.
- Not official TypeSafe.
- Official TypeSafe status, or that jev.pro issues API keys.
- That a schema-constrained answer is automatically factually correct.
Disclaimer
This is an independent unofficial site and is not affiliated with TypeSafe AI; official documentation is available at https://docs.typesafe.ai.
Primary documentation: https://docs.typesafe.ai. Hub: Use cases.
Sources
Public TypeSafe or adjacent documentation only. No private claims.