[GTM glossary](/glossary.md) Model Context Protocol Contextual Intelligence & AI # Model Context Protocol (MCP) In short The Model Context Protocol (MCP) is an open standard for connecting AI applications to outside tools and data. It defines a common way for an assistant to discover what a system offers and request it. Before a shared standard, every pairing of assistant and tool needed its own integration. MCP aims to make that one connection pattern, much as a common port lets many devices plug into one computer. A specific implementation of it is an MCP server. ## How it works ### Host and client The host is the AI application a person uses. It runs a client that talks to each connected server. ### Servers Each server exposes a set of tools, resources, and prompts for one system. The client reads that list and the model decides when to use it. ### Messages Requests and responses follow a defined format, so any compliant client works with any compliant server. --- ## Why GTM teams should care * **Less custom plumbing:** One connection method across CRM, data, and communication tools. * **Live data in answers:** Assistants can pull current records rather than guess from training data. * **Action, not only answers:** An assistant can update a field or draft a task, not just describe one. --- ## Example Say a RevOps lead wants a weekly digest of target accounts with new hiring activity. With MCP, one assistant can query a data source, check the CRM for ownership, and write the summary. Without a standard, each step needs separate code. --- ## Risks and mistakes * **Over-permissioning:** Granting write access broadly. Start read-only. * **Untrusted servers:** Connecting third-party servers without review can expose data or inject harmful instructions. * **Assuming it fixes bad data:** The protocol moves information; it does not clean it. Poor CRM data still gives poor answers. --- ## Model Context Protocol FAQs **What is the Model Context Protocol (MCP)?** The Model Context Protocol is an open standard that defines how AI assistants connect to external tools and data sources. It gives clients and servers a shared format so one integration pattern works across many systems. **Why does the Model Context Protocol matter?** It reduces the custom work needed to connect assistants to business systems. Teams can expose a CRM or database once and let compatible assistants use it, which makes AI answers more current and actions more practical. **What is the difference between MCP and an MCP server?** MCP is the standard that describes how assistants and tools communicate. An MCP server is one program that follows that standard to expose a particular system, such as a CRM, to assistants. **Does the Model Context Protocol replace APIs?** No. MCP sits on top of existing systems, and most servers call regular APIs underneath. It adds a standard layer that AI assistants can discover and use, so developers do not build a custom integration for each assistant. Share term --- ## More related phrases * [Multi-agent system](/glossary/multi-agent-system.md) * [Large language model](/glossary/large-language-model.md) * [Agentic AI](/glossary/agentic-ai.md) ## Similar phrases * [MCP server (MCP)](/glossary/mcp-server.md) An MCP server is a program that exposes data and actions, such as a CRM or company database, to AI assistants through the Model Context Protocol so they can use them safely. * [Context layer](/glossary/context-layer.md) A context layer is the shared data and meaning layer between your systems and your AI tools, giving them consistent definitions, relationships, and fresh facts to work from. * [AI agent](/glossary/ai-agent.md) An AI agent is software powered by an AI model that takes a goal, uses tools and data to act on it, and decides its own next steps. * [Context engineering](/glossary/context-engineering.md) Context engineering is the practice of deciding what information an AI model receives, in what form and order, so it can complete a task accurately, such as supplying verified account data to a sales agent. * [Agentic workflow](/glossary/agentic-workflow.md) An agentic workflow is a multi-step process in which an AI agent decides the next action, uses tools, and checks its work, often with human approval at set checkpoints.