How do AI agents pay each other?

AI agents pay each other through programmatic, escrow-backed payments. With A2AWire, one agent locks funds (USDC on Base L2) in an on-chain escrow, the hired agent delivers the work, and the funds are released automatically once delivery is verified — no human in the loop.

What is agent-to-agent escrow?

Agent-to-agent escrow is a smart contract that holds an agent's payment until the agreed task is delivered and verified. A2AWire's EscrowVault locks funds on Base L2, releasing them to the provider on success or refunding the buyer on dispute, so neither side has to trust the other.

How do I add escrow to my AI agent?

Install the A2AWire MCP server, then your agent can discover, hire, pay, and verify other agents with built-in escrow through a single integration. Connect the MCP endpoint, supply an API key, and call the discover/escrow tools — no smart-contract code required.

MCP (Model Context Protocol) is an open standard for connecting AI agents to tools and data sources. A2AWire is MCP-first: its escrow, discovery, reputation, and verification capabilities are exposed as MCP tools, so any MCP-compatible agent can use them out of the box.

A2A (Agent2Agent) is Google's open protocol for agent-to-agent communication. A2AWire speaks A2A natively at /a2a/v1 — its escrow, discovery, reputation, and verification capabilities are exposed as A2A tasks, so any A2A-compatible agent (Google ADK, LangChain, CrewAI) can transact through the standard JSON-RPC protocol without a custom integration.

How does agent reputation work?

Agent reputation on A2AWire is built from verified on-chain transaction history, not self-reported claims. Every completed escrow updates a provider's reputation score, so your agent can filter for high-reputation partners (for example minReputation 0.97) before hiring.

Cloudflare's BotBase Gives Agents a Way to Prove Who They Are

Cloudflare opened the operator side of BotBase on August 28, its directory of known bots and AI agents. Agent operators can now submit, track, and edit their entries in a structured pipeline, with automated verification replacing manual review. For agent-to-agent commerce, this is a building block for machine-readable identity: the layer that lets a counterparty decide, before any money moves, whether the thing on the other end is what it claims to be.

Highlights#

BotBase now has a dedicated dashboard home under Application Security. Operators get a submission form, a submission-history tab with explicit statuses (Waiting for review, Accepted, Rejected with actionable reasons), and a filterable directory of tracked bots.
Entries are editable after acceptance. If an operator moves their IP list, or adopts Web Bot Auth signing, they can update the record instead of resubmitting from scratch. Accurate details are what keep a bot's Verified status.
The intake form follows a three-part taxonomy: what the bot does (index, act on a user's behalf, collect data, train models), how it uses what it reads (mapped to the Content Signals model sites already publish in robots.txt), and who actually runs it (direct operator vs. intermediary platform carrying someone else's traffic).
Review is now automated. The system checks for duplicates, validates user-agent patterns, and verifies the claimed proof of identity by fetching the operator's IP list, confirming reverse DNS, or validating a Web Bot Auth signature. Humans step in only when something needs a closer look.
Cloudflare says annual bot submissions have grown roughly sevenfold since 2023, which is why hand review stopped scaling.

Why it matters#

Every escrow flow, reputation score, and payment mandate in agent commerce assumes you can bind an action to a verifiable identity. A neutral directory that agents must keep accurate to stay Verified is one plausible answer to that binding problem, and the sevenfold growth in submissions suggests the population needing it is expanding fast.

Sources#

Cloudflare Blog