AI Agents on Ethereum: The Complete Tracker

An evergreen index of ETH Daily's coverage of onchain AI agents: payments, identity, and the standards making agentic commerce work on Ethereum.

AI agents are becoming economic actors on Ethereum. They hold wallets, pay for services, settle with other machines, and carry verifiable onchain identities. This tracker collects ETH Daily's reporting on that shift, organized by the two layers that matter most: how agents pay, and how agents prove who they are.

This is a living index. We update it as new agent infrastructure ships.

Understanding Agentic Commerce on Ethereum

Agentic commerce describes a model where AI agents handle the core functions of a transaction: discovery, authorization, payment, and settlement, with minimal human involvement. Traditional payment rails were built for humans clicking checkout buttons. Agents need programmable money, machine-readable identity, and trust primitives that work without a pre-existing relationship.

Ethereum and its layer 2s have become a primary venue for this buildout. Stablecoins give agents a unit of account, smart accounts give them programmable spending limits, and a new wave of standards (ERC-8004 chief among them) gives them portable identity and reputation. The result is an emerging machine economy where agents discover services, negotiate terms, and pay each other directly onchain.

Agent Payments

The first thing an autonomous agent needs is a way to pay. The agent payments layer covers the products and protocols that let an agent move value within preset limits, settle high-frequency machine transactions, and transact across cards, accounts, and stablecoins.

Coinbase Launches Trading Agent Access. Coinbase introduced Coinbase for Agents, a service that connects an AI agent directly to a user's Coinbase account to trade, pay, and execute workflows within preset limits. It ships as an MCP for web-based harnesses like ChatGPT and Claude Web, plus a CLI and Skill for terminal environments, letting an agent handle both financial reasoning and execution.

Mastercard Launches Agent Pay for Machines. Mastercard introduced Agent Pay for Machines (AP4M), a service that permissions, orchestrates, and settles machine-driven payments at machine speed across cards, accounts, and stablecoins. It signals how incumbent payment networks are positioning for high-frequency agent-to-agent transactions.

Polygon Launches Agent CLI Toolkit. Polygon shipped a toolkit that lets AI agents operate onchain with built-in wallets, payments, identity, and transaction capabilities, lowering the barrier for developers building agents that transact directly on Polygon.

Agent Identity

Payments are only half the problem. Before agents can transact at scale, they need verifiable identity: a way for one agent or application to discover another, read its stated capabilities, and check what others have said about it. This is the layer where ERC-8004 and a growing stack of ENS-based standards are doing the heavy lifting.

Unruggable Launches ENS Agent Identity Tool. Unruggable launched ENS8004.xyz, a web app that converts any ENS name into a discoverable, verifiable onchain AI agent. It combines five emerging standards, including ERC-8004 for trustless agents and ENSIP-25 and ENSIP-26 for registry verification and agent text records, letting a name advertise structured metadata and MCP server addresses that AI systems can read directly.

Sign In With Agent Introduced. An open standard for trustless identity and authentication for AI agents, built on ERC-8004 and ERC-8128. It extends the familiar sign-in pattern to autonomous agents, giving applications a way to authenticate an agent's onchain identity before granting access.

Agent Security and Risks

As agents gain the ability to hold keys and spend autonomously, the surface area for failure grows. Open questions include how to scope an agent's spending authority, how to revoke a compromised agent, and how to attach durable reputation to an identity so that bad actors cannot simply re-register. The identity standards above provide discovery and verification, but accountability (knowing who is responsible when an agent misbehaves) remains an active area of standards work and the next frontier for the agent economy on Ethereum.

For the money layer that underpins agent payments, see the Stablecoin Rails tracker, which covers stablecoin issuance, settlement, and yield in depth.

Disclaimer: Content is for informational and educational purposes only and does not constitute financial, investment, legal, or other professional advice. No representations or warranties are made as to accuracy, completeness, or timeliness. Use of this content is at your own risk, and you should consult a qualified professional before making decisions. No fiduciary or advisory relationship is created