Use cases· Last updated 2026-09-21

HR confidence thresholds with Jev

Inbound applications and tickets mix eligibility, policy questions, and messy free text. The ATS, payroll, and your employment counsel still own outcomes. Jev can score job-related criteria you wrote. It is not an ATS and must not become a proxy for protected-class decisions.

This unofficial page is the confidence thresholds slice of the HR screening operations pack. Intent: apply the Jev (TypeSafe System One) decision model to HR screening operations confidence thresholds. Primary search language: HR 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): Job-criteria Nouls with an explicit do-not-encode-protected-class lint — not a clone of an ATS or HR-screening product page.

HR use-case context

Thresholds turn HR screening operations 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 HR screening operations hides calibration.

Hub: Use cases. Compare, when the other tool is the real job: HR screening.

Confidence Thresholds inputs

You need (1) pinned answers on a frozen contract and (2) labels for screen gold from recruiters, work-auth gold, and whether a human would have held docs. State shape:

{
  "app": { "id": "c-190", "text": "5 years Python; no work auth mentioned; relocating July." },
  "role": { "must": "Python in production; authorized to work without sponsorship this cycle." },
  "ticket": { "type": "inbound_app" }
}

Decision signals and actions

FLOORS = {
    "tag_screen_only": 0.55,      # illustrations — replace
    "auto_reject": 0.95,
}
NOUL_TAU = 0.80  # for work_auth_clear

def allow(ans, action):
    return ans.confidence >= FLOORS[action]

Do not treat a Noul of 0.5 as a “medium” HR screening operations 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 HR screening operations, treat auto_reject as the high bar (auto-rejecting a candidate or taking employment action). Tune on labels — see offline evaluation.

Band around 0.5 on work_auth_clear always reviews. Do not copy 0.80 onto Choice confidence.

Evaluation and rollout notes

• False auto-reject on a planted qualified canary (block ship)
• Missed hold_docs
• Override rate by recruiter — high override means criteria are wrong

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

FAQ

Should auto_reject 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 HR pack? Start with HR decision workflow and HR human handoff. Cluster hub: Use cases.

Is this an ATS or hiring decision system of record? No. It is an unofficial typed-screening pattern. Employment law stays with you and counsel.

Can we score “culture fit”? We recommend you do not. Keep criteria job-related and lint for proxies.

What this page does not claim

• Not an ATS, not employment-law advice, not a fairness certificate.
• No quality-of-hire numbers.
• 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.