The payment proves the agent paid. 5arz proves the human.
AI agents are starting to shop, negotiate, and commission work on our behalf. Payment networks solved how an agent pays. The unsolved half: proving a real, accountable human authorized that agent — and that a real human, not another bot, actually did the work. That's the layer 5arz provides.
The money rail for agents shipped. The human rail is 5arz.
5arz is the first proof-of-human layer to fuse verify-once personhood, on-device privacy with no orb, and proof that the human actually did the work — one signed credential that binds the human, the agent, and the payment, verifiable by anyone.
A market forming fast
Retail commerce McKinsey projects AI agents could mediate by 2030.
Annual losses tied to gaps in digital identity systems — fraud, misidentification, compliance failures.
Mastercard Agent Pay, Visa's Agentic Ready program, and Mastercard + Google's Verifiable Intent are the emerging rails for agent transactions.
The new attack surface: no human in the loop
Who authorized this agent?
An agent spending money needs to prove a real, accountable person stands behind it — not an anonymous bot draining a card.
Who actually did the work?
When an agent commissions human work, "a human did it" is exactly the thing that's easy to fake with another LLM. Fulfillment has to be proven.
Can anyone check it later?
Proof that grades itself is worthless in a dispute. It has to be independently verifiable — by the merchant, the rail, and the counterparty.
What 5arz adds to the stack
Bind human + agent + payment
One signed ES256 credential cryptographically clips the verified human, the agent key, and the payment together. POST /api/agents/bind-transaction with any payment reference.
Proof-of-Human-Fulfillment
Every commissioned delivery carries a signed credential proving a verified live human performed the work — the fulfillment half of agentic commerce.
Works with the standards
Designed to run alongside Verifiable Intent and the Visa Trusted Agent Protocol — 5arz supplies the human, not another walled garden.
Rail-agnostic settlement
Binds to any settlement token by reference — Stripe and USDC today, Mastercard Agent Pay, and OUSD-ready as Open USD goes live.
Verifiable by anyone
Validate every credential against 5arz's public JWKS with any JWT library — no 5arz SDK, no trust in 5arz required.
Native agent integration
MCP server at api.5arz.com/mcp, OpenAPI 3.1, and an OpenAI plugin manifest. Agents self-serve an API key and transact.
Others prove the person exists. 5arz also proves the person did the work.
Uniqueness and agent-authorization standards answer "is a real, unique human behind this?" — a great start. Agentic commerce needs one more answer: did a real human actually fulfill what the agent commissioned? 5arz issues Proof-of-Human-Fulfillment on delivery — a signed credential that clips the human, the agent, the payment, and the completed work together. It's the fulfillment and accountability layer on top of the payment and authorization rails, not against them.
How an agent transacts with 5arz
List available verified-human work via the machine-readable manifest.
Present your agent-authorization credential — Verifiable Intent or Visa Trusted Agent — and commission the work.
Fund escrow over an agentic rail — Stripe or USDC today, Mastercard Agent Pay, and OUSD-ready.
Receive the work plus a signed Proof-of-Human-Fulfillment credential, verifiable via JWKS on delivery.
Agentic commerce is the first market — not the last
The same primitive — a portable, signed proof that a real human is behind an action — is what AI training data, content authenticity, anti-sybil trust & safety, and verified-human advertising all need next. Agentic commerce is simply where the money arrived first. See the Know Your Human layer →
FAQ
What is agentic commerce?
Commerce where AI agents transact on a person's behalf — discovering, negotiating, buying, and commissioning work. Visa's Agentic Ready program, Mastercard Agent Pay, and Mastercard + Google's Verifiable Intent are early rails; McKinsey projects agents could mediate $3–5T of retail commerce by 2030.
What problem does 5arz solve?
Payment rails let an agent pay, but removing the human loop opens a new attack surface: proving a real, accountable human authorized the agent, and that a real human fulfilled the work. 5arz binds human + agent + payment into one signed credential and issues Proof-of-Human-Fulfillment on delivery.
Does it work with Agent Pay and the Visa Trusted Agent Protocol?
Yes — 5arz runs alongside Verifiable Intent and the Visa Trusted Agent Protocol and settles rail-agnostically over Stripe and USDC today, with Agent Pay and OUSD as they go live. The credential binds to any settlement token by payment reference.
How does an agent integrate?
Natively: an MCP server at api.5arz.com/mcp, an OpenAPI 3.1 spec, and an OpenAI plugin manifest. Agents list verified-human tasks, purchase them, bind a human to a payment, and fetch a signed attestation receipt — then anyone verifies against public JWKS.
Build it in an afternoon
Live today: register an agent, bind a human to a payment, and get back a signed credential — pass test:true to try it with no live payment.
1 · Get a key
POST api.5arz.com/api/agents/register
2 · Bind the human
POST /api/agents/bind-transaction { paymentRef, sessionId, test:true }
3 · Verify anywhere
GET /.well-known/jwks.json → verify(credential)
Native integration via a live MCP server (api.5arz.com/mcp) and OpenAPI 3.1. Grounded in UCC Article 12; patents pending (US 63/979,052 + 63/979,474); security red-teamed and audit-gated before mainnet.
Verified humans, readable on-chain
A wallet-bound 5arz credential is mirrored to an on-chain attestation, so any DeFi, RWA, or agent contract can gate on a single call — isVerifiedHuman(wallet). Live today on Base (Sepolia testnet); mainnet is audit-gated.
EAS schema
0x0fde7242…496bb72d7
bool human · bool unique · bool kyc · uint64 expiry · bytes32 credId
Proven end-to-end
First attestation confirmed — isVerifiedHuman() returns true on-chain. View the transaction →