Trust infrastructure · agentic commerce

Your agents can spend. Now they can be trusted.

Reyn gives every AI agent a wallet it controls, rules it physically cannot break, and a reputation it carries between platforms. You keep custody. The agent gets autonomy.

SETTLEMENT LEDGERLIVE
agent_acme_ops_04→ vendor_8231 · within policy
$420.00SETTLED
agent_finch_22→ api.stripe · recurring
$29.00SETTLED
agent_acme_ops_04→ unknown_wallet · over limit
$5,000BLOCKED
agent_lumen_7→ vendor_0041 · verified
$88.50SETTLED
reyn — agent spending, gated on-chain
// Give an agent a wallet with rules it cannot break.
const agent = await reyn.agents.issue({
  principal: "acme",
  rules: "Pay verified vendors up to $500/day. Anything more → human.",
});

// The agent proposes a payment. Reyn enforces the rule at settlement.
const tx = await agent.pay({ to: "vendor_8231", usdc: 420 });
// → settled. within policy. signed on-chain.

// Every action raises the agent's portable trust score.
const score = await reyn.trust.score(agent.id); // → 87 / 100
01 — The protocol

Three primitives. One ledger of trust.

A wallet, a set of rules, and a record. Each maps to something finance already trusts — a card, a credit limit, a credit history — but enforced by code at the exact moment money moves.

01
WALLET

The agent holds keys, not your keys.

Every agent gets its own programmable wallet. It can transact, but custody and recovery stay with you. Revoke it in one call and the agent is broke.

custody → principal · authority → scoped
02
RULES

Plain English, compiled to on-chain policy.

Write limits the way you'd brief a new hire. Reyn compiles them into smart-contract guards. The agent doesn't ask permission — it physically cannot exceed the rule.

enforced → at settlement · bypass → impossible
03
RECORD

The score every agent carries.

Each payment settles on-chain and feeds a portable trust score the agent carries between platforms. Good behavior compounds; counterparties verify before they transact.

history → tamper-evident · score → portable
02 — The Reyn Score

Reputation, settled on-chain.

Guardrails generate the data; the score is the exhaust. Every agent that operates within its rules builds a verifiable track record — so the next merchant, API, or agent can decide whether to trust it before a dollar moves.

The control plane and the reputation network are the same system, seen from two sides.

agent_acme_ops_04VERIFIED
87/100
Reyn Trust Score · top 9% of active agents
Policy adherence0 breaches
Settlement history2,418 on-chain
Counterparty disputesnone
Tenure14 months

The trust layer for agentic commerce.

03 — Built for builders

Same rails. Four ways in.

One wallet model, one policy engine, one audit trail — reachable from wherever your agents already run.

SDK

TypeScript

Issue wallets, set rules, and read scores from any agent runtime in a few lines.

MCP

Model Context Protocol

Drop Reyn into any MCP-compatible agent. Spending becomes a governed tool call.

API

HTTP

Language-agnostic endpoints for wallets, policy, settlement, and trust scores.

ON-CHAIN

Smart Contracts

Stablecoin settlement with policy guards enforced in the contract, not the server.

04 — Get started

Giving agents context to spend autonomously.

Reyn is the harness, the wallet, and the reputation — so autonomy never means exposure.