How **MCP server** is used in this channel In the captured material, an MCP server is the concrete thing you build to put the Model Context Protocol into practice: it takes a website or web service and exposes it as a set of tools that an AI can use natively to get work done. [^ea9761fa-0184-555e-be30-4928a6682680@11] The framing is that the internet's architecture was built for a pre-AI world, and MCP — realized through MCP servers — is the bridge that lets AI hook into it. [^ea9761fa-0184-555e-be30-4928a6682680@0] Who is building them The conversation features Stainless, the API company behind the APIs and SDKs for major AI companies like OpenAI and Anthropic, which is now starting to build MCP servers as well. [^ea9761fa-0184-555e-be30-4928a6682680@22] [^ea9761fa-0184-555e-be30-4928a6682680@34] Open questions The material does not treat MCP servers as a solved technology. The host and Alex Rattray explicitly explore: “So, Alex and I get into the nitty-gritty of what the future of MCP looks like, how to design good MCPs, why MCPs are actually really hard to scale and possibly insecure, and we try to figure out together what a better model for allowing AIs to use the internet might look like.” Takeaways - An MCP server turns a website or web service into a set of tools an AI can use natively. [^ea9761fa-0184-555e-be30-4928a6682680@11] - They are the practical building block of the MCP bridge between the pre-AI internet and AI agents. [^ea9761fa-0184-555e-be30-4928a6682680@0] - Stainless — the API company for OpenAI and Anthropic — is now building MCP servers. [^ea9761fa-0184-555e-be30-4928a6682680@22] [^ea9761fa-0184-555e-be30-4928a6682680@34] - Scaling MCPs is described as hard and possibly insecure, and a better model remains an open question. [^ea9761fa-0184-555e-be30-4928a6682680@46]