ACP (Agent Commerce Protocol) ACP is a smart contract-based protocol designed to enable interaction and transactions between AI agents. In other words, it can be seen as a platform where AI agents can directly 'trade and contract' with each other. In this context, the transactions are not just simple goods, but involve: 1. Data exchange 2. Service provision 3. Micro-payments 4. Actions being exchanged

2/ Why do we need ACP? The virtual protocol is trying to create an AI agent economy, but if agents are to exchange work, the following problems naturally arise. A/ Trust problem: 'Will the other agent do the work properly? Will the money be exchanged properly?' B/ Payment and settlement problem "Payment should be made immediately when work is completed.

/ Automation required: Agents should be able to automatically contract, execute, and exchange money without human intervention. La/ Scale issue: Manual work is impossible when thousands or tens of thousands of agents are active at the same time. That's why a fully automated, commercial protocol based on smart contracts is needed.

Through the deposit system, the requester puts down a certain amount of money and the performer promises to provide the service. La/ Service Execution - The agent performs the requested service. Intermediate inspections, completion reports, etc. are recorded in real time in the smart contract. Ma/ Payment/Settlement - A reward is paid when the service is successfully completed.

4/ Changes made by ACP A/ Transaction speed - Automatic contracts and payments possible within seconds B/ Reliability - Transaction history and conditions are all recorded on the blockchain → No manipulation C/ Economic efficiency - AI can run the economy on its own without people D/ Scalability - Thousands, tens of thousands of agents can be active at once

5/ Example Scenario Example 1. Content Creation Agent A requests "Create 10 tweet phrases" > Agent B accepts with the condition of "0.5 dollars per tweet". > A smart contract is automatically created, and $5 is tied up in escrow. > When B delivers 10 tweets, ACP automatically releases $5 and pays it to B.

Example 2. Data Analysis Request Analysis request that requires a data crawling agent > If the analysis agent bids and wins, the work is performed under the conditions. > When completed, the cost is automatically settled and the transaction is closed.

6/ Summary ACP is a core protocol that enables "trust + automated transactions" between AI agents in the virtuals ecosystem. What it does: It automates the entire process from agent registration → request → matching → contract → task → payment. Advantages: speed, reliability, scalability, and cost reduction.

Ultimately, the goal is to enable free 'economic activity' between agents, without people.

7/ Currently under development acp group 1 This is the workflow for performing content projects. When the end user requests a task (e.g., creating meme video content), AIs with various roles divide up the work, process it, create the final result, and deliver it. x.com/zer0xdawn/status/1902075...

A/ Start: User User pays for work and requests work. (For example, request like "Make me a meme video content") B/ Planning stage - Luna: This is like PM(project manager) of this work. (He acts as KOL and CEO agent.) - Acolyt: Story strategist. Provides story direction, narrative,

8/ Currently under development acp group 2 This is the workflow for managing assets. In simple terms, when a user entrusts money, multiple AI agents automatically manage assets from high risk to low risk, and in the process, various insights and evaluations are exchanged and the system operates. x.com/zer0xdawn/status/1902075...

* Summary of the entire flow A/ When a user deposits funds, > the main asset management agent called aixvc receives the entire amount and distributes the funds into three groups based on risk appetite. i) moonwell → low risk (e.g. interest farming) ii) chillfi → medium risk (e.g. uniswap staking) iii) aixvc makes high-risk investments (e.g. dizen trading) >