Ethereum will implement the ERC-8004 standard on its mainnet, introducing a formal on-chain framework for AI agents to have identities, build reputation, and validate interactions. The goal of this change is to enable inter-agent trading.
The Ethereum Foundation decided to add the ERC-8004 standard to its mainnet, enabling AI agents to trade with each other, building identity and reputation, and increasing interactions within the network.
The proposal emerged from a cross-sector effort and is now positioned as a core element in the Ethereum Foundation’s dAI team’s 2026 roadmap, underscoring its strategic intent for decentralized AI on Ethereum. ERC-8004 defines three lightweight on-chain ledgers that can operate on the Ethereum mainnet or Layer-2 networks.
An Identity Ledger issues a persistent and portable on-chain identifier so that agents can prove their identity across different platforms. A Reputation Ledger is key for network users, as it allows them to review an agent’s history and build security and trust based on it.
Finally, a Validation Ledger authenticates claims and transactions, creating a layer that guarantees agent behavior and data integrity.
Taken together, these records enable discovery, interaction, and payment flows between agents. The protocol leverages HTTP 402 semantics and stablecoin settlement to allow agents to pay for APIs, data, or compute without requiring traditional API keys or accounts, according to project documents and initial reports.
When does the rollout begin within the Ethereum network?
The mainnet activation on January 29 will mark the first full-scale live test of ERC-8004. If successful, it will not only benefit those using AI agents for their operations but also create a more predictable market by allowing for more accurate forecasting of each agent’s actions.
Technically, the most immediate challenge is scaling the validation registry. The protocol’s success hinges on how quickly this layer can process and verify a large number of agent interactions without incurring excessive latency or gas costs.
Developers and infrastructure providers will be pressured to design Layer-2 implementations or clustering techniques that preserve completion and performance.
This launch will reveal whether portable identities and on-chain reputation can sustain real-world market activity across AI agents, and whether the validation mechanisms can scale without undermining cost or performance assumptions—factors that will determine the adoption and economics of a trustless AI agent market.
