What is the Model Context Protocol (MCP)?
Definition
The Model Context Protocol (MCP) is an open standard for connecting AI applications to external data and tools. Introduced by Anthropic in November 2024, it defines how an MCP client inside an AI application talks to MCP servers that expose tools, resources and prompts, so one integration works across many AI products.
Also called: MCP
Updated
Before MCP, connecting an AI assistant to a company system meant writing a bespoke integration for each pairing of assistant and data source. Ten tools and ten systems meant up to a hundred integrations. MCP replaces that with one protocol: each system exposes an MCP server once, and every MCP-capable AI application can use it.
How it works
MCP follows a client-server model. The host is the AI application a person uses, such as a chat assistant or a code editor. Inside it, an MCP client holds a connection to each MCP server. Messages use JSON-RPC 2.0, and servers can run locally over standard input and output or remotely over HTTP.
A server offers three kinds of capability:
- Tools: actions the model can call, such as searching records or creating a task.
- Resources: data the application can read into context, such as files or database rows.
- Prompts: reusable templates a user can invoke.
The model decides when to call a tool; the host decides what the user must approve. Authentication and access control are the server’s responsibility, which is why the quality of an MCP server matters as much as the protocol. See MCP server.
Why it matters for company context
MCP turns context into something portable. If a company’s decisions and commitments sit behind one MCP server, every assistant that speaks MCP can use them, and replacing one assistant with another does not mean rebuilding the integration. That is the argument for building an AI context layer on an open standard rather than a proprietary connector.
MCP vs function calling
Function calling is a model feature: the model returns a structured request to run a named function. MCP is a protocol around it that standardises how those functions, and the data behind them, are described, discovered and served. Most MCP clients use the model’s function calling under the hood.
How In Parallel approaches it
In Parallel exposes its shared context through an MCP server, so Claude, ChatGPT, Copilot, Cursor and other MCP clients read the same decisions, risks and open questions captured from meetings. Each request is answered with what the connected user is allowed to see. The Own your context page explains why the open standard matters for lock-in.
FAQ
Common questions
- Who maintains MCP?
- Anthropic introduced MCP and published it as an open specification with open-source SDKs. It has since been adopted by other AI vendors and developer tools, which is what makes it useful: a server written once can be used from many clients.
- Is MCP only for Claude?
- No. MCP is vendor-neutral. Clients exist in many AI assistants, coding tools and agent frameworks, and any of them can connect to any compliant MCP server.
- How is MCP different from a normal API?
- An API is specific to one service and each AI product needs its own integration with it. MCP is a common protocol on top: the server describes its tools and data in a standard way, and any MCP-capable AI can discover and use them without custom code.
Start with your next meeting.
Bring your team in. Use every surface, including every other team's. From €29 per user per month, billed monthly, no commitment.
30-minute demo. We'll show you how meeting signals become a shared record of decisions, owners and risks.