gatekeeperops
13 days ago
We published a minimal “Proof Pack” for operational AI claims: download bundle + run a read-only verifier → PASS/FAIL.
Includes: decision receipts, append-only audit chain, ROI ledger (baseline vs actual vs counterfactual), deterministic replay.
Release: https://github.com/Sc691583/gatekeeper-proofpack-demo/releas... Spec: https://github.com/Sc691583/gatekeeper-proof-spec Benchmark: https://github.com/Sc691583/gatekeeper-proof-benchmark Contact: info@dayandrossglobal.com
gatekeeperops
13 days ago
Quick start: 1) Download Release 2) Run: python3 verifier.py dist/<bundle>.tar.gz → PASS/FAIL + audit-chain head
gatekeeperops
13 days ago
What this is for: making operational AI claims reproducible and auditable. A “Proof Pack” is a bundle of artifacts (receipts, ROI ledger, audit chain) that can be verified offline.
Happy to answer: determinism strategy, receipt schema, audit-chain format, and how counterfactual ROI is computed.
gatekeeperops
12 days ago
One question I’m curious about: What would you consider the minimum “proof bar” for operational AI in production? We went with: determinism + receipts + tamper-evident audit chain + ROI ledger.