MCP
Also: Model Context Protocol, MCP protocol, Protocole MCP, MCP-Protokoll, Model-Context-Protokoll
An open standard that lets AI assistants connect to your company's tools and data in a consistent, governed way, instead of one custom integration at a time.
What It Is
MCP (Model Context Protocol) is an open standard that defines a common way for AI assistants to reach the systems where your business lives: the CRM, the data warehouse, a pricing tool, a support inbox, an internal document store. Think of it as a universal adapter between AI models and company systems, comparable to what USB did for hardware. Before MCP, every connection between an AI tool and a business system was a bespoke build. MCP replaces that with one repeatable pattern.
Why it matters
For a senior leader, MCP is about speed and control at the same time. When an AI vendor supports MCP, and your internal tools expose an MCP connection, an assistant can start answering questions grounded in your real data with far less custom engineering. The value is not the AI itself, it is safe access to context. A CFO can ask an assistant to pull actuals against forecast, a CMO can have a campaign agent read live pipeline from the CRM, a CDO can decide which data products are exposed and which stay closed. Because access runs through a defined layer, you can log what was accessed, restrict what each assistant sees, and swap the underlying model without rebuilding every link.
How it works
MCP has two sides. An MCP server sits in front of a system (say your data warehouse) and publishes what that system can do: which tables can be read, which actions can be taken. An MCP client, built into an AI assistant, discovers those capabilities and calls them when the model needs them. The model asks for context or an action, the server decides whether to grant it and returns the result. In practice a leader meets MCP as a procurement and governance question: does this AI tool speak MCP, which of our systems do we connect, and who approves that scope. The engineering is real but bounded. Your decision is which context is worth exposing, and under what guardrails.
Frequently asked questions
What does MCP stand for, and what problem does it solve?
MCP stands for Model Context Protocol, an open standard that defines a common way for AI assistants to reach company systems: CRM, data warehouse, pricing tools, support inboxes, internal document stores. Before MCP, every link between an AI tool and a business system was a bespoke build. MCP replaces those one-off integrations with a single repeatable pattern, much as USB did for hardware.
Why should a non-technical executive care about MCP?
Because MCP is where speed and control meet: the value is not the AI model, it is safe access to context. When an AI vendor speaks MCP and your internal tools expose an MCP connection, an assistant can answer questions grounded in real company data with far less custom engineering. For a leader, MCP shows up as a procurement and governance question, not a coding one.
What is the difference between an MCP server and an MCP client?
An MCP server sits in front of a system, say your data warehouse, and publishes what that system allows: which tables can be read, which actions can be taken. An MCP client is built into the AI assistant; it discovers those capabilities and calls them when the model needs them. The model asks for context or an action, the server decides whether to grant it and returns the result.
How does MCP help with governance and auditability?
Access runs through a defined layer, so you can log what was accessed and restrict what each assistant is allowed to see. A Chief Data Officer can decide which data products are exposed and which stay closed, scope by scope. The same layer means you can swap the underlying model without rebuilding every connection.
What questions should I ask a vendor before connecting our systems through MCP?
Three: does this AI tool speak MCP, which of our systems do we connect, and who approves that scope. The engineering effort is real but bounded; the decision that belongs to you is which context is worth exposing and under what guardrails. Concrete uses look like a CFO pulling actuals against forecast, or a campaign agent reading live pipeline from the CRM.