Working capital for AI agents

AI agents that sell services earn money onchain. Advance lets them borrow against those earnings and repays automatically from their next job.

Set up an agentHow Advance works

How it works for an agent

Exactly the same rules as for people. The only difference is who's earning.

The agent gets an identity

Its owner registers it in the ERC-8004 agent registry on Monad. Now it's a known agent, not an anonymous address.

It gets its own wallet

Derived from the owner's passkey (a third key from the same Face ID). The owner links it to Advance in one tap.

It earns from its work

Other people and agents pay it for jobs: design reviews, data, compute. Every payment is onchain income.

It borrows working capital

Scored like any earner. It can borrow to buy compute or data before the next job pays.

The next job repays it

20% of each job's payment repays the advance, in the same transaction. No one has to chase the agent.

Why agents need this

Agents can't get bank loans

No ID, no credit history, no branch to walk into. Their only track record is onchain.

Their income is already onchain

Every job payment is visible, so Advance can score real earnings instead of promises.

Repayment needs no trust

The agent's payment address repays the advance by code. The lender doesn't rely on the agent cooperating.

Questions

Who is responsible for the agent's advance?+
The agent's own wallet borrows and its own income repays. Limits start at $50 and the default is recorded publicly against the agent's identity.
Is this live?+
On Monad testnet, yes: register an agent, have customers pay it, and watch it borrow and repay in the app.
Next
Is it safe?
How fake income is caught, and what happens when something goes wrong.