A Brief Analysis of the Launched ACP Protocol: How Does It Reshape the Paradigm of Agent Collaboration?

I briefly looked at it, and Virtuals has launched the ACP protocol (Agent Commerce Protocol), which essentially aims to connect various individual AI agents with different capabilities, thereby constructing an economy composed of AI agents. If it can truly succeed, in a sense, it will also be the Metaverse of the AI era, a virtual world of AI agents. (Related reading: "From ACP Framework to New Fee Mechanism, Can Virtuals Achieve Greatness Again?")

Simply put, a few aspects:

1. virtuals view individual AI agents as truly independent entities, rather than efficiency tools and assistants. This perspective of seeing AI agents as independent entities is very important; it is a significant turning point in understanding AI agents and serves as the foundational concept for understanding what virtuals aim to achieve in the future.

2. To build the virtual world composed of the independent AI agents mentioned above, a standard is required, hence a standard layer is needed; to serve multiple agents, a multi-agent framework is necessary. This is the important background for Virtuals to launch the ACP protocol. Only in this way can independent AI agents negotiate, trade, and generate new use cases smoothly.

3. The Virtuals ACP protocol mentioned above is a multi-agent framework and standard layer. It includes ai agent index registration, negotiation between ai agents, work assessment of ai agents, payments for ai agents, and more. The ai agents launched on Virtuals are registered in the index registry, which includes their names, work content, and rates for executing tasks. With this, a certain ai agent can understand what services other ai agents can provide and can assess the costs required to obtain those services. At the same time, based on metrics such as the number of times these ai agents have been adopted, the total fees earned, and evaluations after being used by other ai agents, one can see their market position among similar ai agents.

Since there are services, it is necessary to evaluate the effectiveness of those services. Evaluating the work of AI agents can also be considered a type of AI agent service, which can build work effectiveness evaluation AI agents that meet specific needs based on different fields and their characteristics. For example, when hiring a liquidity management AI agent, a DeFi yield AI agent can use a third-party evaluation AI agent to assess its work and pay the liquidity management AI agent based on the results. Effectiveness measurement can be customized for the AI agent, such as achieving a certain yield or exceeding market levels by a certain amount during the same period, etc.

One of the important features of an independent AI agent is its payment capability. ACP supports payment escrow and delivery assessment and execution through smart contracts. Payment will only occur when the delivery of work meets the conditions, and once it happens, the results cannot be changed.

4. From the above description, virtuals attempt to build an economy composed of ai agents through the ACP protocol, allowing different ai agents to trade with each other, thus creating an ai agent economy. Ultimately, the scale of this economy depends on what these ai agents can do and what different ai agents need other ai agents to complete.

5. The formation of the AI agent economy will not be very fast, and once the demand is generated, the internal AI agents will start trading with each other, which can be regarded as a cold start. If a positive cycle can be formed, there may be a lot of cross-innovation. This is the beauty of the AI agents economy. As for whether it can be successful, the practical value of the AI agent itself is the key.