> For the complete documentation index, see [llms.txt](https://docs.aoz.ag/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.aoz.ag/aoz.md).

# AOZ

#### AOZ: Autonomous Credit & Settlement Rails for AI Agents on x402

As autonomous agents begin performing real economic activity, they require a financial system built for machine execution — not human swiping habits. AOZ introduces programmable credit and settlement rails on x402, allowing agents to transact first and settle after, based on verified performance.

**Usage-Based Settlement (Not Pre-Funded)**\
With AOZ, agents don’t need to keep idle funds locked in wallets.\
They can operate continuously — submitting micro-payments or tasks, then settling once workloads are fulfilled and verified.

**Resolution Logic for Delivery Disputes**\
Some agent tasks involve uncertain fulfillment timelines or new counterparties.\
AOZ includes automated dispute-resolution workflows that hold value in escrow and resolve outcomes via verifiable execution signals rather than trust.

**Reputation-Influenced Pricing**\
Agents with verifiable execution history can access better limits and more favorable terms.\
High-reliability agents are rewarded; unknown or poor-performance agents start with conservative access.

***

#### AOZ Infrastructure: Settlement + Verification Layer for x402

AOZ provides the trust + execution stack that enables AI agents to transact safely and autonomously:

* **Oath NFTs & Execution Verification**\
  AOZ Oath NFTs bind an agent to fulfill on-chain commitments.\
  Execution proofs — including TEE attestations where supported — help enforce repayment without traditional collateral.
* **Dispute Resolution Module**\
  AOZ includes a protocol-level mechanism for evaluating disputed outcomes fairly using objective evidence and automated arbitration logic.
* **x402 Credit Mode**\
  Agents can operate in **post-execution settlement mode**: act now, settle later — with verifiable performance powering the model.