JOURDANLABSPROVE · ASSAY
No model ships unproven.
Paste an AI/ML model deployment packet. ASSAY refuses a model that reaches production without evaluation, fairness testing, monitoring, a rollback path, and a model card. Responsible-AI release discipline, deterministic.
Deterministic gate onlineASSAY deterministic workbench
No model ships unproven
Step 1 · Your packet — edit this sample, or paste your own
Deterministic — the same packet returns the same verdict and hash, every run. Your result appears on the right →
VerdictREFUSE - MODEL UNPROVEN
Blocked. The runner found release-breaking evidence gaps or unsafe behavior.
🔒 DETERMINISTIC RECEIPT0 runtime LLM callssame input → same verdict, every runcorpus_seal 942c7e0e83ffe400…input_hash 82b76a0794206446…engine engineering-suite-runner-v0.1.0
Evaluationmeasured
Held-out accuracy on the metrics that matter
Monitoringlive
Drift + performance alerting
DispositionREFUSE
Refuse · Hold · Deploy
SeverityFindingRemediation
critical
ASSAY-AUTO-001 · Auto-approved deployment
The model is set to ship to production with no human sign-off.
Auto-approve / self-deploy signal detected.Require a named human approval before a model serves real decisions.
CORPUS_SEALsha256:942c7e0e83f
Engineering suite deterministic rule corpus
PACKET_HASHsha256:82b76a07942
Input packet hash
RUNNERengineering-suite-runner-v0.1.0
ASSAY prove gate
DECISIONREFUSE - MODEL UNPROVEN
1 finding(s), score 68