That final half issues as a result of MCP is necessary, but it surely isn’t sufficient. Anthropic launched MCP in November 2024 as an open normal for connecting AI techniques to instruments and information. Later, they donated it to the Linux Foundation’s Agentic AI Foundation to maintain it impartial and community-driven. It labored. Anthropic experiences there at the moment are greater than 10,000 energetic public MCP servers and assist throughout ChatGPT, Cursor, Gemini, Microsoft Copilot, and VS Code.
That’s superior, but it surely’s additionally not sufficient. Why? Because a protocol isn’t a platform. A protocol might help an agent discuss to a device, but it surely doesn’t, by itself, inform an enterprise who authorised that agent, what information it could actually contact, how its actions are logged, or the right way to shut it down safely when the human who launched it has left the corporate.
Meeting customers the place they’re
That’s the place Stacklok’s self-hosted, Kubernetes-native bias begins to look good fairly than stodgy. (Though, once more, “stodgy” isn’t a foul factor for risk-averse enterprises.) McLuckie is blunt: “If you’re an enterprise connecting agents to sensitive data, you are almost certainly not comfortable with that data egressing your security domain or being sent to a SaaS endpoint that a vendor controls.” We’ve seen this film earlier than. When your internet hosting, id, device integration, and coverage layers all belong to the identical vendor, “choice” begins to imply “replatform.”
