AI Agent's New Exploration in the Web3 Field: From Manus to MCP

Recently, a global first universal AI Agent product named Manus has attracted widespread attention. As an AI tool capable of independent thinking, planning, and executing complex tasks, Manus demonstrates unprecedented versatility and effectiveness, providing new ideas and inspiration for the development of AI Agents.

AI Agent, as an important branch of artificial intelligence, is gradually transitioning from theory to practice and demonstrating great potential across various industries, with the Web3 industry being no exception. The core components of AI Agent include large language models (LLM), observation and perception mechanisms, reasoning and thinking processes, action execution, and memory retrieval.

Currently, the design patterns of AI Agents mainly have two development routes: one focuses on planning ability, while the other emphasizes reflective ability. Among them, the ReAct pattern is the most widely used design pattern, and its typical process includes three steps: thinking, acting, and observing, forming a cyclic iterative process.

Based on the number of agents, AI Agents can be divided into Single Agent and Multi Agent. Single Agent mainly focuses on the combination of LLM and tools, while Multi Agent completes complex tasks through collaboration among agents with different role positioning.

Model Context Protocol (MCP) is an open-source protocol launched by Anthropic, aimed at addressing the connectivity and interaction issues between LLMs and external data sources. MCP provides three capabilities: knowledge augmentation, function execution, and pre-written prompt templates, utilizing a Client-Server architecture and based on the JSON-RPC protocol.

In the Web3 industry, despite a significant decline in the market capitalization of AI Agent-related projects, some projects remain active. These projects are primarily divided into three categories: the launch platform model represented by Virtuals Protocol, the DAO model represented by ElizaOS, and the business company model represented by Swarms.

From the perspective of economic models, currently only the launch platform model can achieve a self-sustaining economic closed loop. However, this model also faces the problem of the assets themselves lacking attractiveness, as many launched AI Agents essentially remain Memes that lack intrinsic value.

The emergence of MCP has brought new exploration directions for AI Agents in Web3. One approach is to deploy the MCP Server on a blockchain network, addressing single point issues and possessing censorship resistance; another is to endow the MCP Server with the ability to interact with the blockchain, lowering the technical barrier. In addition, there are plans for a creator incentive network based on Ethereum called OpenMCP.Network.

Although the combination of MCP and Web3 can theoretically inject decentralized trust mechanisms and economic incentives into AI Agent applications, there are still some limitations in the current technology, such as the difficulty of verifying the authenticity of Agent behavior with zero-knowledge proof technology, as well as the efficiency issues of decentralized networks.

The integration of AI and Web3 is an inevitable trend. Although there are still many challenges at present, we need to remain patient and confident, continuously exploring the development possibilities in this field. With technological advancements and the emergence of more innovative applications, the prospects for AI Agents in the Web3 ecosystem will be even broader.