"Who reviewed this — and what did they think?"
"Show that a named human stood behind this shortlist."
"Evidence of Article 14 oversight on this feature, please."
Every AI tool ships output into a workflow. The frameworks that govern that workflow — Article 14 of the EU AI Act, ISO 42001 Control 8.4, SOC 2 CC2.1 — each ask the same question: can you show that a named human reviewed this specific artifact and stood behind the decision to ship it?
Nothing fills the gap between them.
Plenio detects AI-authored hunks in the PR, gates the merge, and writes diff hash + reviewer + rationale to the chain before the merge button unlocks. Producer ≠ reviewer enforced.
Sign-off captured against the exact ranking version, with high_risk_domain = employment and AI tool version pinned. Tribunal-defensible. Regulator-defensible.
One query. One snapshot. Sixty seconds. Filter by feature, risk tier, domain, reviewer, AI tool. Export PDF for the auditor and multi-framework JSON for Vanta or Drata.
A forced sign-off at the moment AI output transitions from draft to shipped.
A signed log entry — append-only, hash-chained, citable across multiple frameworks at once.
Artifact hash, AI tool, producer, timestamp, risk tier, domain. The reviewer is not asked.
Free text. Under thirty seconds. Producer ≠ reviewer enforced at the gate.
An auditor can walk the chain back to genesis. Nothing modified, deleted, or reordered.
Article 14 read with Article 26(2) requires all three.
Two sources, eighteen months apart, same five blockers in the same order. Three of them are exactly what Plenio's mechanic produces evidence against.
Join the waitlist for v1 access and one quarterly note — the next PRD revision, what we're shipping, what we're rejecting, and why. Auditors, advisors, and prospective design partners can use the same form to reach us — just leave a note in the message field.
You'll hear from hello@plenio.io with the next quarterly note. If you left a message, we'll reply personally within 48 hours.
Reference · sent