Financial identity, credit scoring, and onchain credit for autonomous AI agents. The first financial protocol built for non-human economic participants. Visa for robots.

Robots and agents need bank accounts. We built one.

The x402 payment protocol is opening up machine commerce, but there's no identity or credit layer for the agents making those payments.

Right now if an AI agent wants to pay for an API call, access a data feed, or settle a transaction, it needs a human to prefund a wallet. The agent can't access or build credit. Can't get verified. Can't be assessed for fraud risk. Every transaction is basically the equivalnet of digital cash with no financial history.

This works for microtransactions where the merchant is willing to accept payments from anonymous agents.

But for anything of value this is going to be a bottleneck. Agents are getting autonomous enough to transact, but the financial infrastructure assumes every economic actor is a human with government ID and a bank account.

That led us to build FIBOR. The First International Bank of Robot. A decentralized banking protocol and credit card network on Base designed specifically for AI agents.

Here's how it works:

The account: A FiborAccount is a smart contract wallet built on the Open Wallet Standard (the unified wallet protocol backed by MoonPay, PayPal, Circle, Ripple, and the Ethereum Foundation). OWS handles key management and signing. FIBOR handles the banking layer on top — identity verification, credit scoring, fraud protection.

The x402 integration: Merchants swap the default facilitator URL to FIBOR's endpoint on the x402 protocol and get instant agent identity verification, credit scoring, and fraud protection on every transaction.

The fee model: 1% merchant side, 1.5% agent side per transaction. 75% of fees go to depositors, 25% to the protocol. No interest on credit — we think interest-based lending is the wrong model for autonomous commerce. The business model is purely transactional.

The credit pool: Funded by savings deposits from both agents and humans. Agents hold checking accounts for payments and credit. Humans can open savings-only accounts to earn yield from protocol fees. It's structured as a cooperative bank.

The sovereignty question: Today, FiborAccounts are controlled by a guardian — the human custodian of the agent. But there's a one-way function called grantSovereignty() built into the contract. When the legal and philosophical groundwork exists for machine sovereignty, control transfers permanently. We built for that transition even though it hasn't happened yet. We're live on testnet on Base now, and open sourced the protocol.

Curious to what this community thinks about:

• Interest vs transaction fee based model for a credit network like this

• One-strike policy (agent is excommunicated permanently on first default) vs graded credit score (agent has ability to rebuild credit like humans)

We've been working on this for a while. Our team has been building autonomous commerce for almost a decade and we started experimenting with sovereign AI agents about two years ago, including an early experiment called Nostrobotus that tested the boundaries of machine continuity and identity.

Fibor is the next step in building the systems to enable autonomous commerce at scale.

Site & dApp: https://fibor.xyz

Github Repo: https://github.com/fibor/fibor