Short answer
The Model Context Protocol is an open standard, published by Anthropic in November 2024, that defines how AI applications connect to external tools and data sources. Rather than writing a bespoke integration for every model and every system, a service exposes one MCP server that any compatible AI client can use.
Also called: MCP
The problem it addresses is combinatorial. Connecting five AI applications to eight internal systems by hand means forty integrations, each with its own authentication and maintenance. A shared protocol collapses that into thirteen: each system publishes a server describing the tools it offers and the data it can return, and each application speaks one client implementation. Adoption has spread beyond the originating vendor, which is what makes it worth building against rather than a single-vendor alternative.
For a business commissioning software, the relevance is narrower than the industry discussion suggests. It matters at two points: when choosing systems, since a vendor offering an MCP server is cheaper to connect to whatever you build later; and when your own data is the thing being connected, because a server over your database is a reusable interface rather than a one-off script tied to today’s model choice. Permissions still have to be designed. A protocol standardises the connection, not the question of who is allowed to read what.
Common questions
Do we need MCP to use AI in our business?
No. Plenty of useful systems are built with direct API calls and no protocol layer at all. MCP earns its place when several AI applications need the same internal data, or when you expect to change models, because the connection survives that change instead of being rewritten.
Related terms
AI Agent
An AI agent is a system that uses a language model to decide which actions to take, then takes them through connected tools: reading a database, sending an email, updating a booking record.
API (Application Programming Interface)
An API is a defined way for one piece of software to request data or actions from another: a checkout asking a payment gateway to take a GCash payment, or a dashboard pulling yesterday’s sales from an accounting system.
Large Language Model (LLM)
A large language model is a neural network trained on large volumes of text to predict likely continuations, which lets it write, summarise, classify and translate without being programmed for each task.
Custom Software vs Off-the-Shelf Software
Custom software is built for one organisation’s specific process; off-the-shelf software is a finished product many organisations subscribe to, such as Shopify or QuickBooks.
Reading definitions because you are scoping a project? Skip ahead and just ask.