ERC-8004: The Future of AI on Ethereum

The Ethereum ERC-8004 update is here: the new standard that promises a new lease of life for AI agents.

Agent-based artificial intelligence represents one of the fastest-growing developments in the current global technology landscape.

Unlike a traditional Large Language Model (LLM), which merely generates a response based on an input, an AI agent is designed to pursue objectives autonomously, breaking complex problems into subtasks and using external tools to improve its behaviour based on the results. 

Thanks to AI agents, we are essentially moving rapidly from purely reactive artificial intelligence systems to systems capable of acting in the real world with a significant degree of autonomy.

However, this expansion has not yet been matched by the development of adequate shared infrastructure underpinning relationships among the various agents, with the current system forcing agents to operate in fragmented environments, lacking standardised mechanisms to verify the identity of counterparties or to access portable reputation data.

ERC-8004 aims to bridge this gap by defining an open standard for the identity, reputation and validation of AI agents on the Ethereum blockchain. 

If you’re curious to find out what the ERC-8004 update is and how it could position Ethereum as the fundamental trust layer upon which the future of distributed artificial intelligence rests, you’ve come to the right place.

The problem of trust between agents

Imagine a future where an AI agent manages your DeFi portfolio: monitoring prices, moving liquidity between protocols, evaluating returns, and executing trades autonomously. Now imagine that an agent needs to hire a second, more specialised agent — perhaps one that calculates a protocol’s risk, or one that optimises transaction fees. 

At this point, a series of questions might arise, such as: how does the first agent know whether it can trust the second? How does it verify its credentials? How does it check its performance history? How does it ensure it isn’t interacting with an imposter?

Until now, there has been no shared, open answer to these questions. Each platform adopted proprietary solutions, and trust between agents depended on centralised intermediaries.

ERC-8004 stems from a proposal released on 13 August 2025, which subsequently went live on the mainnet on 29 January 2026, by four of the most prominent developers in the Ethereum ecosystem — Marco De Rossi, Davide Crapis, Jordan Ellis and Erik Reppel — to resolve precisely this problem through a system comprising three ledgers, recreating a history of an AI agent’s behaviour within the blockchain.

What is an ERC

Before delving into the more technical aspects of this article and understanding what ERC-8004 is and how it works, it is useful to take a step back and clarify what is meant by the term ERC.

ERC stands for Ethereum Request for Comments, a set of technical documents that define shared standards for the operation of certain components of the Ethereum ecosystem.

Essentially, an ERC represents a formal proposal through which developers establish common rules that anyone can adopt — just like the protocols that allow browsers and servers to communicate with one another, regardless of who built them.

When the community adopts a standard, it becomes a fundamental building block of the blockchain infrastructure, upon which developers can build, knowing that their products will be compatible with the rest of the ecosystem.

An ERC is therefore an open convention, proposed by developers, discussed publicly and adopted voluntarily to create a more standardised and efficient technological environment.

How ERC-8004 works: the three registries

The ERC-8004 standard, designed to become a benchmark for AI agent infrastructure within the Ethereum ecosystem, achieves this standardisation by establishing three key registries: the ‘identity registry’, the ‘reputation registry’ and the ‘validation registry’.

Each of these registries intervenes at a specific stage of an agent’s lifecycle, thereby creating a verifiable, interoperable and trustless system. Let us now look more closely at each registry to understand in detail what ERC-8004 is and how it works.

The first registry, the identity registry, fulfils the critical function of anchoring the agent on the blockchain, representing its formal and unique identity.

In a purely digital and potentially ephemeral context, this identity is consolidated through the issuance of a non-fungible token (NFT) compliant with the ERC-721 standard. This technical choice not only guarantees authenticity but also enables the transferability of the agent’s ownership and granular management of access rights.

The informational core of this registry lies in the so-called Agent Card, a JSON document that serves as a universal technical specification. Within it, operational endpoints, declared capabilities and payment addresses are mapped; this allows any smart contract or user to uniquely identify the methods of interaction with the synthetic entity, without resorting to centralised intermediaries.

The second registry, the reputation registry, constitutes the agent’s operational history, acting as an immutable archive of behavioural and technical performance. In this second pillar, trust is quantified through the on-chain recording of structured feedback, generated at the end of each interaction by users or other coordinating agents. 

This architecture transforms qualitative data into a transparent empirical metric that reflects critical parameters such as execution reliability, response latency, and the accuracy of the outputs provided.

To achieve this objective, the registry introduces a mechanism based on applied game theory: since building a positive reputation requires a constant commitment of resources and time (opportunity cost), the accumulated reputational capital acts as collateral against malicious behaviour. In the absence of a central rating authority, the Reputation Registry enables selecting agents solely on cryptographic evidence, making operational merit a verifiable, immutable digital asset.

The third and final component, the validation registry, introduces a level of external verification necessary to mitigate the information asymmetries inherent in artificial intelligence models. 

This registry enables qualified third parties — such as security auditors, academic institutions, or agents specialising in validation — to issue formal attestations regarding specific competencies, security standards, or the agent’s ethical compliance.

The combination of these three elements represents a true paradigm shift in how AI agents interact with one another: for the first time, identity, reputation, and validation are brought together and transformed into shared, portable resources that anyone on the blockchain can verify.

Why it matters for the future of Ethereum

Ethereum’s history is written in its standards. These are not merely technical documents, but the foundations upon which entire cycles of economic innovation rest. This happened in 2015 with ERC-20, which gave rise to the token industry, and in 2017 with ERC-721, which opened the doors to the world of NFTs, transforming both cases into global markets.

ERC-8004 follows exactly in this vein, but raises the stakes further by applying the logic of standardisation to agentic artificial intelligence.

With this standard, Ethereum ceases to be merely a network for financial transactions and positions itself to become the universal trust layer for autonomous agents, thereby creating the technological environment necessary for the growth of an on-chain synthetic economy, where complex services are delivered and verified without human intervention.

Understanding what ERC-8004 is and how it works means having the tools to grasp the next phase of Ethereum’s maturity: the one in which the programmability of value finally meets the autonomy of AI, transforming the network’s use into concrete, measurable utility rooted in the real economy.