Decision Passport logoDecision PassportVerifiable AI governance

Architecture

Off-chain evidence, on-chain decision passport

Decision Passport uses Solana as a verifiable governance layer. The chain proves integrity and lifecycle state, while sensitive evidence stays in controlled off-chain systems.

1

Agentic AI workflow

Produces a structured recommendation with risk, confidence, model version, policy version, and a non-sensitive evidence summary.

2

Off-chain evidence vault

Keeps prompts, logs, code, PII, customer names, and sensitive evidence outside the chain and outside the public verifier.

3

Canonical hash service

Normalizes JSON deterministically, hashes the decision and evidence bundle, and creates the privacy-preserving passport.

4

Solana passport record

Stores hashes, schema version, issuer, reviewer, status, timestamps, and supersession or revocation pointers only.

5

Public verification

Recomputes the decision hash and proves PASS or FAIL without disclosing sensitive evidence.

What stays off-chain

  • Prompts, model transcripts, raw logs, and source code.
  • PII, customer names, vulnerability details, and private evidence.
  • Large supporting documents and reviewer workpapers.

What the passport stores

  • Decision hash and evidence bundle hash.
  • Model version hash, policy version hash, and schema version.
  • Issuer, reviewer, status, timestamps, revocation, and supersession pointers.

Why the blockchain is useful here

Independent verification

A stakeholder can prove whether a decision object still matches its anchored hash.

Lifecycle integrity

Approvals, revocations, and supersessions become explicit status transitions.

Privacy boundary

The public record is useful because it is narrow, not because it exposes everything.