This holiday, I focused on learning about MCP.
Currently, MCP is not as good as many people say. The future should be bright, but there is still a long way to go.
- MCP is not very friendly for beginners. While the configuration has become much simpler, it still requires some basic knowledge, and there are significant differences in some MCP configuration methods.
In this regard, we can look forward to products that allow beginners to use MCP without much thought, such as simply checking a box to enable the desired MCP services. Additionally, it should allow saving usage plans, etc.
- Currently, MCP is only suitable for self-experimentation to improve personal efficiency; it cannot be shared or used with others' plans. Therefore, it cannot be used commercially at this time.
In fact, MCP is a type of AI Agent solution, which is definitely much simpler compared to using Dify workflow methods. Many scenarios now have packaged AI Agents, such as Manus or those in Doubao. Beginners are better off using ready-made solutions.
- MCP consumes tokens relatively quickly, and the response time can be slower, which does not meet the needs of time-sensitive scenarios. For example, it is not suitable for my current Jin.
免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。
