The Model Context Protocol, or MCP, is an open standard that lets an AI assistant plug into outside tools, like a CRM or a calendar, through one shared kind of connector.
Anthropic introduced MCP on November 25, 2024. It called MCP "an open standard that enables developers to build secure, two-way connections between their data sources and AI-powered tools." MCP gives every tool the same way to connect, so an assistant can reach your calendar, files and other software without a separate build.
One connector works across Claude, ChatGPT and other AI tools
MCP's official website describes it as "an open-source standard for connecting AI applications to external systems." It adds, "Think of MCP like a USB-C port for AI applications." USB-C cables work with phones, laptops and monitors from different makers, and your CRM's MCP connector can work the same way across different AI apps. The website lists Claude, ChatGPT, Visual Studio Code and Cursor, and that flexibility matters if you want to switch AI providers or use more than one.
OpenAI's developer docs describe MCP as "an open protocol that's becoming the industry standard for extending AI models with additional tools and knowledge." OpenAI also notes that remote MCP servers can connect models over the internet to new data sources and capabilities. For a small business, this means a connector built for one AI app can often work with another app too. No rebuild needed. It lowers the risk of getting stuck with a single AI vendor.
How MCP connects an agent to your business tools
MCP connects three pieces to turn a request into a completed action. Each piece plays a role.
- The AI app. This is the assistant your team talks to, like Claude or ChatGPT.
- An MCP server for each tool. Your CRM, calendar or file storage gets its own connector, built to the MCP standard described above.
- The tool request. When the agent asks to use a tool, the MCP specification requires that hosts "must obtain explicit user consent before invoking any tool." The AI app checks with you first, before anything happens.
Anthropic opened MCP to a wider foundation
Anthropic created MCP. The protocol no longer belongs to one company. The company donated MCP to the Agentic AI Foundation, a directed fund under the Linux Foundation. Anthropic, Block and OpenAI co-founded the fund, with support from Google, Microsoft, Amazon Web Services, Cloudflare and Bloomberg, according to Anthropic. For a small-business owner, that backing is a sign the standard has broad industry support.
Connected tools can take real action, so consent comes first
A connected tool can do more than answer a question, since it can add a calendar entry, update a CRM record or send a file. For a business, this could mean the difference between an agent answering a question about a customer, and reaching out to that customer directly. Because of that, the MCP specification states that "users must explicitly consent to and understand all data access and operations" before they happen. The specification also warns that "tools represent arbitrary code execution and must be treated with appropriate caution." MCP itself can't enforce these rules at the protocol level, according to the specification. Responsibility falls to whoever builds the connection. That's the team setting up your agent.
Questions to ask before connecting an agent to your tools
Our team put together this list from the consent rule at the center of the MCP specification. Ask these questions before you connect any agent, no matter which AI tool your business uses.
- Ask which tools the agent can reach, and whether that list can grow later.
- Ask whether the agent must get your approval before it takes an action, like sending an email or changing a record.
- Ask who reviews and maintains the connector for each tool over time.
- Ask what data the agent can see, and whether that matches the job it's doing.
- Ask how you would find out if something went wrong after a tool took action.
For help connecting an agent to your tools, see our AI systems and agents service.
Frequently asked questions
What is MCP in simple terms? MCP is an open standard that lets an AI assistant plug into outside tools through one shared kind of connector. The guide compares it to a USB-C port for AI applications.
Can MCP connect tools across different AI apps? Yes. The guide says a connector built for one AI app can often work with another app too, and lists Claude, ChatGPT, Visual Studio Code, and Cursor.
Should an AI agent ask before using my tools? Yes. The guide says the MCP specification requires hosts to get explicit user consent before invoking any tool, and that users must consent to and understand data access and operations before they happen.
