Behind the x402 craze, how ERC-8004 enables AI agents to have "credit scores"

This article delves into the ERC-8004 standard, which aims to establish a decentralized trust layer for AI entities, providing verifiable identities and reputation records, addressing trust issues that the x402 protocol cannot handle independently. (Background: The Ethereum Foundation's dAI team released the 2026 roadmap: Decentralized AI agents focus on ERC-8004 and x402) (Background supplement: What is the Ethereum ERC-8004 protocol? Creating “on-chain identity cards” for AI agents: Endorsed by Google, MetaMask, and Coinbase) Recently, with the x402 protocol designed for automatic micropayments between AI entities (Agent) becoming a market hotspot, discussions about its early potential have been nonstop. However, beneath the aura of x402, a more fundamental standard—ERC-8004—quietly enters the sight of keen observers. A new question arises: Since x402 has already aimed to solve payment issues, what is the positioning of ERC-8004? Are the two in competition? The answer seems to be negative. They are more like a pair of intricately designed, complementary components in the business landscape of AI entities. Source: A post on X by Davide Crapis, head of AI at the Ethereum Foundation. One detail worth noting is that the creator of x402, Erik Reppel from Coinbase, is also one of the final signatories of the ERC-8004 protocol. Among those who co-signed this standard are core representatives from MetaMask, Google, and the Ethereum Foundation. This heavyweight lineup and high personnel overlap strongly suggest that the two are not in competition but are collaboratively constructing a grander blueprint. The logic of this blueprint is very clear: If the emergence of x402 validates the immense demand for AI agent payments, then ERC-8004 represents another core element necessary for building this vast machine economy. What it targets is precisely the core problem that x402 cannot solve independently—where does trust come from in an economy of autonomous AI entities? The Bitget Wallet Research Institute may guide you to find the answer in this article. The “trust” issue of entity collaboration: The dilemma that x402 cannot solve. In the business collaboration of AI entities, payments (the problem solved by x402) are merely the last step of a closed loop. A more fundamental question precedes it: How can an AI entity be sure that another AI entity it hires is trustworthy? In other words, how can entities prove their capability to complete the tasks delivered? To answer this question, one must explore the foundational framework of entity business. According to the theory supported by Davide Crapis of the Ethereum Foundation, a decentralized AI entity business system must rely on three pillars to operate stably: Discovery (Discovery), Communication (Communication), and Computation (Computation). The three pillars of the entity business ecosystem consist of: Core Function Technical Implementation Role and Description Discovery Layer Entity Discovery ERC-8004 constructs the first step of trust and collaboration. ERC-8004 enables AI entities to locate and identify each other through its Ethereum-based identity registry. Communication Layer Agent Communication x402, A2A, AP2 facilitates interaction between entities. x402 is an open standard for on-chain payments, parallel to protocols like Google’s A2A and AP2, handling payments and data exchange between entities. Computation Layer Verifiable Computation EigenCompute, EigenAI ensures that all entity reasoning and actions can be cryptographically verified, with the results recorded in the data availability layer, achieving reliable computation. Source: X tweet forwarded by Davide Crapis. From the above table, it can be seen that the x402 protocol mainly addresses the payment standard issues in the communication layer, but it cannot answer the critical questions of identity verification and reputation assessment. Before making payments, AI entities first need to be securely “discovered” to complete the construction of trust and collaboration—ERC-8004 is precisely the solution to this problem, which is to build a decentralized trust layer. ERC-8004 is led by the Ethereum Foundation's dAI team and Consensys, in collaboration with institutions such as MetaMask, Google, and Coinbase, aiming to become the on-chain “trust layer” for AI agents. It is a decentralized business registration and enrollment system—providing each AI agent with verifiable identities, complete historical performance, and capability proofs. All information will be immutably recorded on-chain and made publicly accessible to all participants. Source: ERC-8004 official website. ERC-8004: The decentralized trust layer for entities. ERC-8004 is positioned as the underlying infrastructure of the entity economy, with its core value being: through blockchain technology, it addresses the fundamental problem of the lack of trustworthy identities among AI entities, freeing them from the constraints of traditional centralized platforms and establishing verifiable, cross-platform collaborative relationships. The operation of this mechanism begins with the input of trust data by ecosystem contributors: AI developers are responsible for registering the unique identities of entities; application developers continuously provide behavioral feedback of entities to accumulate reputation data; and verification services act as independent auditors, providing trustworthy on-chain verification of the operational results of entities. Source: Organized by Bitget Wallet. This information collectively feeds into the ERC-8004 registration center, which consolidates the three core functions of the protocol: Identity Registry (Identity Registry): It grants each AI agent a unique digital identity through the ERC-721 NFT protocol. This design not only makes identities transferable but also allows them to link to standardized “agent cards” (Agent Card), which detail the agent's name, capabilities, and metadata, ultimately achieving permissionless discovery across platforms. Reputation Registry (Reputation Registry): This mechanism acts as a decentralized review system based on identity registration. Clients or other entities can submit structured feedback. Its key design lies in the fact that these evaluations can be linked to x402 payment proofs, ensuring that only genuine transactional participants can evaluate, thus effectively reducing fraudulent activities and ensuring the public transparency of all reputation data. Validation Registry (Validation Registry): It provides the ultimate protection for high-risk or high-value transactions. Entities can request trustworthy verification from third parties, such as TEE oracles, staked inference (staked inference), or zero-knowledge machine learning proofs (ZK-ML proofs). These verification methods can provide cryptographic proof for specific model executions and outputs, thus constructing a verifiable question similar to traditional market professional certification.