NORNR mark NORNR Mandates, approvals and evidence for autonomous agents.

NORNR / Economic Control Infrastructure

Verified live

Give agents a mandate before you give them money.

NORNR sits between intent and settlement. It decides whether an agent may spend, under which mandate, against which counterparty, and leaves a defensible trail behind it.

Policy Policy engine live
Approvals Human review verified
Audit trail Signed receipts on record

Traditional rails answer whether payment is possible. NORNR answers whether the spend should happen at all.

When agents buy APIs, data, compute and services, payment is not the hard part. Governance is.

Legitimacy Teams must know why a spend was allowed, not only that it happened.
Mandate Every agent needs limits, exceptions, approvals and allowed counterparties.
Responsibility Human-in-the-loop is not optional when real money can move.
Evidence Receipts, comments, proof and export history belong in the product, not in a reporting afterthought.

The product is a mandate, an approval state and a receipt trail.

  • It is a control layer between spend intent and settlement.
  • Wallets, balances and rails are support objects, not the product identity.
  • The core objects are mandate, approval state, counterparty review and receipt trail.
Approved Policy satisfied · intent released immediately
Queued Operator review required before settlement
Blocked Rejected by policy · no settlement occurs

A balance records what moved. A trail records why it was allowed.

Intent arrives with purpose, amount and counterparty. Policy decides: approved, queued or blocked. Operators review from the same record. Settlement and evidence stay attached.

02

Approval queue

Human approval stays in the same trail — actor, comment and notification state intact.

Slack · email · in-app · API callback
03

Receipt chain

Evidence, invoice references and notes make the spend defensible.

signed · timestamped · publicly addressable
04

Audit export

Signed manifests and close bundles give finance a clean exit.

JSON · CSV · signed manifest

Live proof already exists in the deployed system, not just in the deck.

Rail, approvals, artifacts and SDK paths are verified against the deployed runtime.

Live transfer / March 15, 2026

Live governed settlement completed

0xf8ff5d333d1bdaf932673314853e02406f1ce3e73e0c782ec5e3c007526fe927

Intent approved, rail executed and locked in the audit trail.

Amount $12.50
Counterparty openai
Mandate Builder / owner-approved
Trail settled · signed · archived
Live approvals

Slack approval flow verified

Queued intent moved through signed Slack interactivity and into settlement pending.

Audit artifacts

R2 archive and signed manifest verified

Artifacts persist to object storage and resolve through artifacts.nornr.com.

Developer path

Python and TypeScript quickstarts verified

Agent registration is now the control surface. Adapters ready for OpenAI Agents SDK and LangChain.

TypeScript
import { Wallet } from "@nornr/sdk";

const wallet = await Wallet.create({
  owner: "research-agent",
  dailyLimit: 100,
  requireApprovalAbove: 20,
  baseUrl: "https://nornr.com",
});

await wallet.pay({ amount: 5, to: "openai", purpose: "model inference" });

Start free. Upgrade when autonomous spend becomes operational.

We are looking for near-production agent workflows with real spend pressure.

Mandate $0 free forever

Register up to 3 agents and $500 governed spend.

  • Policy engine
  • Receipt trail and dashboard
  • Intent log
Audit $299 per month

100 agents, $100k governed and finance-grade close packages.

  • Cost attribution by team and project
  • Shadow policies
  • Signed close bundles
Infrastructure Custom negotiated

Unlimited agents with negotiated limits, retention and compliance posture.

  • Deployment and data residency options
  • SSO / SAML
  • Support and SLA by agreement

Subscriptions carry the product today. Settlement fees only apply when NORNR executes the settlement rail for that spend.

Tell us what your agents need to spend on.

We review applications asynchronously. No call required.

What do the agents need to spend on?
What controls matter most?