OrchestriAI
All terms

Model Context Protocol (MCP)

An open standard for connecting compatible AI applications to tools, resources, and prompts.

Technical details checked 2026-08-11

Concept map

Follow the connections to see how the term fits into a broader automation system.

In one sentence

An open standard for connecting compatible AI applications to tools, resources, and prompts.

Model Context Protocol is an open protocol introduced by Anthropic in 2024 and now hosted by the Linux Foundation's Agentic AI Foundation. It standardizes how compatible AI applications discover and use tools, resources, and prompts exposed by MCP servers.

For businesses, an MCP server can make approved CRM, EHR, accounting, or internal-system capabilities available to an AI host without building a separate proprietary connector for every compatible client. The server defines its capabilities and schemas; the host decides what is exposed to the model and what the user may invoke.

MCP does not automatically make access secure or accurate. Authentication, authorization, consent, rate limits, auditability, validation, and human approval are implementation responsibilities. An MCP server often wraps existing APIs, so the security and data-handling rules of those underlying services still apply.

A server may run locally or remotely. That describes where the server executes, not where tool results are ultimately processed. Review the AI host, model provider, permissions, retention settings, connected vendors, and deployment architecture before exposing sensitive data.

Keep following the system

Questions about how this applies to your business?

Book a call