Deterministic enforcement · Ed25519-sealed receipts

It won't let your AI skip the step that matters.

AgenticRail is a deterministic gate that refuses an AI agent's out-of-order step before the action runs, not after. Every decision it makes, allowed or refused, is signed into a receipt anyone can verify offline, with no callback to us. It doesn't make the AI right. It makes the order unskippable, and any later change to the record detectable. There is no language model anywhere in it — zero AI influence on the gate itself. No chatbot, nothing that guesses. The same request against the same sequence state returns the same verdict every time, which is what makes a decision reproducible by someone else rather than regenerated.

The gap

The harm that leaves no trace

When a human is meant to check what an AI made or drafted, the safeguard is real but the check leaves no record. When something goes wrong, there is nothing to point to.

What it does

Two powers, at the boundary that matters — the moment before an AI acts.

agent  →  gate  →  ALLOW / DENY  →  sealed receipt

The proof — verifiable offline, without an account

Every receipt is signed with Ed25519 and verifiable offline against a public key we publish — no callback to AgenticRail. Re-hash the record and check the signature in your own code, or paste a sequence ID into the report tool. A worked example shows the whole thing done on a real sealed sequence: the chain, one receipt complete, and the code that verifies its signature offline.

No sequence ID yet? Make one in two commands. The docs carry a copy-paste request that runs against the live gate on the public demo key — nothing to install, no sign-up. It returns a sequence ID; one more call returns that sequence's receipt with its Ed25519 signature and the exact preimage to check it against. If you would rather not open a terminal, the browser demo drives the same gate and opens the signed report for your run in one click.

Or wire it into your own code. The SDKs wrap the same public API: pip install agenticrail for Python, with LangGraph and CrewAI integrations, and npm install @agenticrail/core for JavaScript and TypeScript. Both run against the public demo key with no account, and the docs carry the full payload contract.

Signed receipt · illustrative shape
decisionALLOW
stepreview_and_sign
payload_hashsha256 of the exact request
signature_algEd25519
key_idk2_2026-06-07_ed25519
signaturebase64 — verify offline or at /report
sealedtrue

This one is illustrative. Generate a real one from the docs, or in the browser →

The AI drafted it. What proves a human checked it?

The same shape turns up wherever AI now drafts a decision. A scribe writes a clinical note and a clinician must review it before it is saved. A model marks an assessment and a moderator must confirm it. An agent assembles a credit decision and someone must approve it before funds move. In every case the safeguard is real, it is required, and it leaves no evidence it happened.

This is where AgenticRail fits. Placed at the sign-off, it refuses that step until the steps before it have run, and signs a receipt that the sign-off was submitted and permitted, in order, under your key. Put a hash of the artifact and the reviewer's ID in the attestation and they are signed into it too.

It can't force a careful read, and it doesn't claim to.

What it is — and isn't

It isIt is not
Deterministic enforcement — the verdict is computed, not an AI's guessA lie-detector for the AI. It records what was decided and in what order, not that the AI was right
A tamper-evident, offline-verifiable recordA stop on hallucination, or a force on human attention
Metadata and hashes — request inputs never enter the receipt at all, which carries a SHA-256 of the whole payload instead, and whatever you place in attestation is published verbatimA claim on your data or your governance — it is a bounded instrument. We hold the receipt signing keys today, and who holds them is a deployment term

Where it comes from

AgenticRail was derived from whakairo — the carver's discipline, where order is law and the finished form is sealed. That lineage is the reason for the seal, not decoration on it. The whakapapa →

Published briefs & specifications

Written to be cited, fingerprinted to be checked. Each carries a SHA-256 fingerprint you can recompute yourself.

Automated Decisions and the Provable Safeguard — three tiers of safeguard (asserted, enforced, provable), what the difference looked like in Robodebt, and a five-minute test anyone can run.
Customer Due Diligence When an AI Agent Runs the Checks — sector gap analysis: the rules require the checks to happen before the business relationship, and an agent that runs them writes its own record of having done so, in an order it also chose.
MAS SAFR Component Mapping: A Deterministic Pre-Execution Gate — the Monetary Authority of Singapore’s reference architecture for agentic finance, mapped component by component. A reference architecture, not a rule: it confers no status.
AI in NZ Health Care: The Missing Evidence Layer — sector gap analysis: national AI-scribe rollout, the human-review safeguard, and the record that doesn't yet exist.
AI in NZ Education Assessment: The Missing Evidence Layer — sector gap analysis: NCEA authenticity, moderation, and the self-attested evidence chain.
When the Marker Is a Machine — an agreement rate is a property of a system; a challenge is about an instance. What evidence supports one AI-assisted result.
If AI Detectors Don't Work, What Does? — prohibit, detect, or record the process. Why the first two both interrogate the artifact, and what the third one changes.
NCEA Is Being Replaced. What Assures the Internal Assessment? — under NZCE and NZACE internal assessment becomes universal, and its moderation is still unspecified.
The Completeness Specification — eight requirements (R1–R8) that separate an evidence-grade enforcement record from an ordinary log.
The Enforcement Specification — the canonical spec: decision architecture, receipt fields, signing, and the sealed chain. Versioned and frozen.
Writing — technical posts.

Verify a receipt, take a key, or start a conversation.

Evaluation is free and needs no account. A developer key is US$39 a month — your own lane, your own declared step order, and receipts that persist rather than being deleted on the demo lane’s 30-day clock. It unlocks no capability the demo key lacks. A deployment has no list price and starts as a conversation.

What it is  ·  Run the live demo  ·  Verify a sequence  ·  Buy a developer key  ·  Questions  ·  hello@agenticrail.nz

Hokianga Harbour Heads at sunset, looking west through harakeke toward the open Tasman — Northland, Aotearoa New Zealand
Hokianga Harbour Heads, Northland, Aotearoa — where AgenticRail is built

He toi whakairo, he mana tangata