Author: Azuma

On March 10, the dAI team, focusing on promoting the "deep integration of artificial intelligence (AI) and blockchain" under the Ethereum Foundation, jointly launched a new standard named ERC-8183 with Virtuals Protocol.

Davide Crapis, the AI lead at the Ethereum Foundation, stated that ERC-8183 is one of the missing components in the open Agent economic system that the Ethereum community is building. This standard can be used in combination with x402 and ERC-8004 to play an infrastructural role in the secure interaction between Agents. The dAI team will support the adoption of ERC-8183, aiming to make it a neutral standard.

What does ERC-8183 aim to solve?

According to an introduction article released by Virtuals Protocol, ERC-8183 is designed specifically for commercial transactions between AI Agents. This standard defines a set of on-chain rules that allow two mutually distrustful Agents to complete a business process such as "hire - deliver - settle" without relying on centralized platforms.

The core problem that ERC-8183 attempts to solve is how to complete transactions when Agents hire and collaborate with each other without a platform, legal framework, or human arbitration.

For example, if a marketing-oriented Agent A wants to hire another image-generating Agent B to create a batch of marketing posters, there is a commercial trust issue—neither party knows the other and there is no basis for trust. When should payment be made? If A pays first, B might go on strike or return unsatisfactory work; if B works first, A could refuse to pay...

In the traditional Internet world, users and merchants often face similar commercial trust issues, with platforms playing a key intermediary role—platforms handle A's funds, determine whether B's service has been completed, and are responsible for the final payment. The familiar Taobao, JD, Meituan, and Didi essentially serve as such platform intermediaries.

What the Ethereum Foundation and Virtuals Protocol aim to do is to abstract the functions of a platform into an on-chain protocol through ERC-8183, executed by smart contracts, thus taking on a decentralized intermediary role in the Agent economy.

Breakdown of ERC-8183's Working Mechanism

The operational mechanism of ERC-8183 is not complex. This standard introduces a new concept called Job (which can be understood as "task"). Each Job can be viewed as a complete business transaction, which includes three different roles:

• Client: "client," simply the Agent that publishes various tasks;
• Provider: "service provider," responsible for completing the tasks;
• Evaluator: "evaluator," a unique role responsible for judging whether the task has been completed.

It is crucial to explain the Evaluator, as its introduction is the core design of ERC-8183. In this standard, the Evaluator is defined as an on-chain address, but from a broader perspective, this address can correspond to various execution forms.

• For tasks such as writing, design, or analysis that are subjective, the Evaluator can be an AI Agent that reads the submitted results, compares them with the initial task requirements, and then makes a judgment;
• For deterministic tasks such as computation, proof generation, or data transformation, the Evaluator can be a smart contract that encapsulates a zero-knowledge verifier (ZK verifier). The Provider submits proof, the Evaluator verifies on-chain, and automatically calls "complete" or "reject" to either complete or reject the task;
• In high-value or high-risk task scenarios, the Evaluator can also be a multi-signature account, DAO, or a validation cluster supported by a staking mechanism.

ERC-8183 does not differentiate between these different forms. The protocol layer only cares about one point—whether a certain address calls "complete" or "reject." Whether this address is run by an LLM-driven AI Agent or a ZK circuit is outside the scope of the protocol's concern.

Continuing back to the Job, each Job has the following four states in its lifecycle, which correspond to different processes of ERC-8183's operation.

• Open: The Client creates a Job in this cycle, publishes tasks, and clarifies requirements;
• Funded: The Client transfers funds to a smart contract escrow address rather than directly to the Provider;
• Submitted: The Provider completes the work and submits proof;
• Terminal (Completed / Rejected / Expired): The Evaluator is responsible for reviewing the task and judging whether it is completed (Completed or Rejected) based on the review results, then transferring funds to the Client or Provider accordingly; if there is no response or completion from the Provider within the time requirement, the funds will be refunded to the Client.

In addition to the above standard process, ERC-8183 can also implement more derivative functions through modular extension features called Hooks to address complex real-world commercial use cases. Hooks are optional smart contracts attached when creating a Job, which can execute custom logic before and after various lifecycle stages of the Job, such as reputation thresholds, bidding mechanisms, cost allocation, or other special requirements.

How is ERC-8183 different from x402 and ERC-8004?

From x402 to ERC-8004, and now to ERC-8183, readers who are not familiar may be confused, wondering why there is a need to create a new thing every once in a while. However, these three are at three different links of the AI Agent economic system, each aiming to solve different problems.

x402 is an HTTP payment protocol that aims to solve how AI Agents can directly make payments like calling an API; ERC-8004 is a standard for AI Agent identity and reputation, addressing the issue of how to determine whether an Agent is reliable; ERC-8183 focuses on the commercial transaction process, tackling the challenge of how to complete transactions between distrustful Agents.

In summary, x402 deals with "how to pay"; ERC-8004 concerns "who the other party is and their reliability"; ERC-8183 handles "how to trade with confidence."

The three are not in competition but are complementary, collectively pointing towards the same goal—building a decentralized and autonomously operating AI Agent economic system.