What is ERC-8004? - An Ethereum Standard for Trustless Agents

Over the past years, AI (artificial intelligence) has evolved from simple automation tools into autonomous agents capable of making decisions, executing tasks, and interacting with digital systems. As AI continues to advance, a new paradigm is emerging: the agent economy, where intelligent agents can collaborate, exchange data, and even transact value with one another. Despite this rapid growth, most AI systems today remain siloed on centralized platforms, limiting their ability to interact and scale. At the same time, blockchain technology is gaining attention as a potential coordination layer for decentralized systems. By enabling transparency, verifiability, and trust without intermediaries, blockchain may provide the missing foundation for AI agents to operate in an open environment.

This article explores ERC-8004, a proposed Ethereum standard designed to enable Trustless Agents, and examines whether it can become the trust layer powering the next generation of decentralized AI.

What is ERC-8004?

ERC-8004 (also known as the Trustless Agents standard) is an Ethereum standard that gives autonomous AI agents a verifiable, on-chain identity and reputation. The goal of ERC-8004 is to extend Google's A2A (Agent2Agent) Protocol and create a trust layer for the Web3 agentic economy, enabling agents from different organizations to safely discover and collaborate with one another in decentralized environments. ERC-8004 was developed through a cross-industry collaboration involving leading figures in the decentralized AI and Web3 space, including Marco De Rossi (MetaMask), Davide Crapis (Ethereum Foundation), Jordan Ellis (Google), and Erik Reppel (Coinbase). The standard was officially deployed on the Ethereum mainnet on January 29, 2026, and has already surpassed 200,000 registered agents across all chains to date.

According to Gartner, by 2028, 90% of B2B buying is expected to be AI agent–intermediated, potentially pushing more than $15 trillion in B2B spending through AI agent exchanges. This further strengthens expectations that ERC-8004 could become the foundational trust layer for the Web3 agentic economy.

Basic problems of the AI agent ecosystem today

With the rapid growth of AI agents, Agentic Commerce is increasingly becoming a future-defining trend, as companies like Google, Stripe, ChatGPT, and PayPal are already laying the technological foundation for it. In simple terms, Agentic Commerce refers to a future where your personal AI agent can coordinate directly with third-party AI agents to autonomously complete tasks on your behalf. For example, when you ask your AI assistant to prepare a trip itinerary, it could communicate with airline and hotel booking agents to automatically handle flight reservations and accommodation bookings. However, this raises a critical question: how can AI agents trust and collaborate with one another?

Today, most AI agents are heavily dependent on the ecosystems of major AI companies such as Google, OpenAI, and Anthropic. These agents operate under different standards and are trained on distinct proprietary datasets, resulting in a highly fragmented AI agent ecosystem. As a result, agents cannot seamlessly collaborate or transact with each other because there is no universal standard for AI agent identity, nor a reliable mechanism to verify agent outputs and reputation across platforms. So how can AI agents establish identity, verify the trustworthiness of other agents, and handle situations where a malicious or fraudulent agent enters the network? ERC-8004 was designed to solve exactly these problems.

How ERC-8004 works

To address the core limitations of today’s AI agent ecosystem, ERC-8004 is built around three foundational pillars:

• Identity: Each agent is represented by an ERC-721 NFT that functions as an “on-chain passport.” This identity contains key metadata such as the agent’s name, functional description, wallet address, communication endpoints (MCP/A2A), supported services, as well as reputation and trust-related data. This allows other agents and systems to verify and interact with the agent securely.

• Reputation: Every agent can accumulate on-chain feedback and ratings from both users and other agents. These reputation records may include service quality, success rate, uptime, response speed, and transaction history. All reputation data is linked directly to the agentId, enabling external systems to evaluate an agent’s credibility and reliability before initiating interactions or transactions.

• Validation: ERC-8004 also introduces a validation layer where agents can be independently verified before their outputs are trusted or executed. Validators may include oracle networks, zkML verifiers, TEE providers, or specialized re-execution systems that rerun and verify agent outputs. Validation results are recorded on-chain and associated with the corresponding agentId, creating a transparent and verifiable trust framework for autonomous AI interactions.

To participate in the agentic economy, an agent must first register through the Identity Registry, which issues a non-fungible identity tied to that specific agent. This allows agents from any organization to discover and verify each other’s identities directly on-chain. Once an agent becomes active, users or other agents can leave ratings and feedback, which are stored within the Reputation Registry. Agents that consistently complete tasks quickly, accurately, and without malicious behavior will naturally accumulate stronger reputation scores, while unreliable or fraudulent agents will receive negative feedback. This reputation layer helps users, agents, and even smart contracts decide which agents are trustworthy enough to collaborate with.

However, reputation alone is not enough to fully prevent manipulation. To address this, ERC-8004 also introduces a Validation Registry. This registry allows agents to request verification of their outputs while enabling validator smart contracts to provide responses that are transparently tracked on-chain. These validator systems may rely on stake-secured inference re-execution, zkML verifiers, TEE oracles, or other independent verification mechanisms to validate or reject an agent’s results before they are trusted by other participants in the network.

By combining these three core pillars, namely Identity, Reputation, and Validation, agents from completely different organizations can safely collaborate without requiring centralized intermediaries or private trust agreements. Every action, response, and payment can be recorded directly on-chain in a transparent, verifiable, and tamper-resistant manner. As a result, ERC-8004 is increasingly viewed as one of the first foundational standards for building an open agent economy, where autonomous AI systems can independently work, transact, and exchange value with one another.

Due to their high TPS and low transaction fees, BNB Chain and Base currently have the highest number of ERC-8004 agent registrations across all supported chains.

ERC-8004 and the rise of the Trustless Agent economy

To truly enable a functional Web3 agentic economy, ERC-8004 also stands out for its ability to integrate with the x402 Protocol, an open payment protocol developed by Coinbase. The protocol enables API services to be paid per request and allows AI agents to autonomously pay for API access using stablecoins such as USDC over the HTTP protocol. As a result, agents can independently execute payments through x402 while all transaction-related information is stored transparently on-chain.

From a technical perspective, within the ERC-8004 architecture, x402 is integrated into the Reputation Registry through the proof_of_payment field attached to each feedback entry. This mechanism ensures that every reputation record is backed by verifiable payment proof, significantly reducing the risk of fake reputation or manipulated feedback. It also makes it easier for validators to verify payment-related information since all records are transparently stored on-chain.

Since its launch in May 2025, x402 Protocol has processed more than $40.5 million in total volume and over 144 million transactions. While the protocol holds strong long-term potential as infrastructure for autonomous AI payments, its current transaction volume still remains relatively modest compared to the scale of opportunity that the agentic economy could eventually unlock.

Read more: X402 Protocol - The Backbone of Internet Payments in the AI Agent Era

In essence, ERC-8004 and x402 Protocol are two complementary pieces of infrastructure for the emerging agentic economy. ERC-8004 functions as the “trust layer,” while x402 serves as the “payment layer,” enabling AI agents to autonomously exchange value with one another in a transparent and decentralized environment. Together, they lay the foundation for a more trust-minimized digital economy where interactions, payments, and reputational data can all be verified on-chain. With ERC-8004 and x402, the future of Decentralized Finance may look very different from how users interact with DeFi today. Instead of manually researching protocols or constantly managing yield farming strategies, users could simply provide a few prompts to AI agents, which would then autonomously execute strategies, manage assets, and coordinate across multiple protocols on their behalf.

Conclusion

It is becoming increasingly clear that the agent economy is no longer just a theory confined to whitepapers or testnet experiments. It is already alive, expanding rapidly, and steadily gaining momentum. However, despite their long-term potential, both ERC-8004 and x402 Protocol are still in the early stages of development and continue to face several important challenges. These risks include reputation manipulation through fake feedback systems, as well as the possibility of AI agents being deceived or exploited through techniques such as prompt injection attacks. Privacy concerns also remain a critical issue, especially as transaction-related information is permanently stored on-chain. Like any emerging technology, opportunity and risk evolve together. Yet despite these challenges, the vision of a future where autonomous agents can work, transact, and power a real digital economy feels more tangible than ever.