How does the Zero-Risk Enterprise Pilot refund guarantee work?

Qualified enterprise teams receive evaluation and adversarial red-teaming for 50 edge-case model outputs or RAG prompts completely free of charge. If our speed and qualitative precision do not demonstrably exceed your internal benchmarks, no invoice is issued and zero obligation exists.

What happens if an evaluation batch fails the Cohen's Kappa agreement threshold?

JudgeMyAI enforces a strict quality gate of Cohen's κ ≥ 0.85 (Fleiss' κ ≥ 0.82 for multi-annotator pods). Batches falling below κ = 0.80 are halted immediately for calibration. If re-calibration cannot be achieved within 48 hours, the client is issued a 100% pro-rata refund or credit.

Can dedicated evaluator pods be cancelled mid-sprint?

Yes. Dedicated pods (5 to 50+ specialists) can be cancelled or scaled down with a 14-day written notice to legal@judgemyai.com. Any unallocated hours or unexecuted sprint retainers are refunded pro-rata to the original payment method.

What is the policy on delivery delays exceeding SLA limits?

If benchmark audit sets exceed our 48-hour SLA or production batches exceed agreed SOW sprint schedules without mutual written extension, client accounts receive an automatic 15% service credit per 24 hours of delay, up to 100% of the batch value.

COMMERCIAL GOVERNANCE // JAI-REF-2026

Commercial Transparency &
Refund Governance.

Enterprise evaluation contracts require mathematical accountability. JudgeMyAI guarantees empirical quality thresholds, transparent milestone reconciliation, and an unconditional zero-risk pilot protocol.

Apply for AI Jobs

DOCUMENT ID: JAI-EVAL-2026-V2.4

EFFECTIVE DATE: September 18, 2026

GOVERNANCE: Enterprise SOWs & Retainer Pods

CONTACT: legal@judgemyai.com

FRAMEWORK SPECIFICATION // SECTION 9

The Zero-Risk Enterprise Pilot Initiative

100% UNCONDITIONAL AUDIT

To eliminate vendor evaluation risk completely, JudgeMyAI offers qualified AI foundation labs and venture-backed engineering teams an initial evaluation of 50 edge-case model outputs or RAG retrieval prompts completely free of charge. We deliver back a verified audit dataset complete with atomic token-span tags, step-by-step reasoning rationales, and Inter-Annotator Agreement scores.

$0 UPFRONT

Zero Upfront Capital

No credit card, no retainer deposit, and no contractual lock-in required to initialize the 50-task pilot.

48h TURNAROUND

Guaranteed Sprint SLA

Audit datasets delivered within 48 hours of prompt ingestion via our secure enterprise transfer pipeline.

0% OBLIGATION

Zero Invoicing Clause

If our speed, domain precision, and agreement metrics do not demonstrably exceed your internal benchmarks, no invoice is ever issued.

Claim 50-Task Pilot Audit Protected by Strict Mutual 2-Way NDA (contact: legal@judgemyai.com)

MATHEMATICAL ACCOUNTABILITY

Objective SLA Gates & Refund Triggers

We do not measure quality subjectively. Every contract is bound to auditable statistical agreement formulations.

1. Inter-Annotator Agreement (IAA) Quality Gate

Per Section 1.1 of our technical framework, JudgeMyAI enforces an agreement standard of Cohen's κ ≥ 0.85 (and Fleiss' κ ≥ 0.82 for multi-annotator pods).

The κ < 0.80 Remediation Policy:

If any deliverable batch falls below κ = 0.80, production is halted immediately. If our senior lead arbiters cannot calibrate the annotator pool and re-deliver the batch to κ ≥ 0.85 within 48 hours, the client receives an immediate 100% pro-rata refund or credit for the entire batch via enterprise@judgemyai.com.

Continuous statistical tracking across golden calibration sets.
Zero client liability for discarded uncalibrated batches.

2. Delivery Latency SLA Guarantee

Per Section 8.1, standard evaluation turnaround times are strictly governed by batch volume:

AUDIT SPRINT (≤5k tasks)

24–48 Hours

PRODUCTION (50k tasks)

5 Business Days

Delay Credit Schedule:

If a batch exceeds agreed delivery schedules without mutually documented technical blockers, an automatic 15% service credit per 24 hours of delay is applied, up to 100% of the SOW batch fee.

Dedicated Pod Retainers & Milestone Reconciliation

For institutional clients running custom dedicated evaluator pods (5 to 50+ full-time specialists in legal, clinical, or software engineering), contracts operate on structured monthly milestone cycles.

Unallocated Capacity Roll-Over

Unused task credits or evaluator hours accrued during an active contract month automatically roll over for up to 90 days.

14-Day Pod Dissolution

Clients may scale down or dissolve dedicated pods with 14 business days' written notice to legal@judgemyai.com. Remaining unspent retainer fees are refunded in full.

Pro-Rata Wire/ACH Return

Approved refunds are returned to the originating corporate bank account via SWIFT, Fedwire, or ACH within 5 business days via support@judgemyai.com.

PROCUREMENT CLARIFICATIONS

Frequently Asked Questions

What conditions trigger a 100% refund on evaluation batches?

A full pro-rata refund is triggered if a deliverable fails to meet our Inter-Annotator Agreement standard (Cohen's κ ≥ 0.85) and cannot be corrected within 48 hours, or if an audit set experiences an unapproved delay exceeding 5 business days past the agreed SLA date. Contact legal@judgemyai.com to initiate processing.

How do I claim the 50-task Zero-Risk Pilot?

Engineering teams can request pilot onboarding directly via enterprise@judgemyai.com. We execute a mutual 2-way NDA, establish secure data transmission parameters, and deliver audited datasets with reasoning rationales within 48 hours. No payment information is required.

Are customer datasets retained if a refund or cancellation occurs?

Never. Under our enterprise data governance standards, JudgeMyAI never trains commercial foundation models on customer prompts or context. Upon contract cancellation or refund execution, all client data and generated annotation artifacts are permanently purged from isolated evaluation environments per SOC 2 Type II guidelines. Direct compliance verification queries to privacy@judgemyai.com.

What happens if our domain experts disagree with JudgeMyAI evaluators?

Discrepancies exceeding a 10% variance are automatically escalated to Stage 4 Lead QA Arbitration. If a genuine interpretative ambiguity exists, JudgeMyAI performs dual-blind re-annotation with domain specialists (MDs, CPAs, or JDs) at zero additional charge to reach consensus.

Deploy Elite AI Evaluators With Zero Friction.

Audit your frontier model against hallucination and instruction drift with guaranteed empirical precision.