MCP connectors: USB-C for AI

The most common way people use AI is also the most labor-intensive: find something, copy it into the chat, get an answer, copy the result back out. Most of the time doesn't go into thinking — it goes into shuttling data back and forth. MCP removes that middle step. It's an open standard — the Model Context Protocol — that works like USB-C for AI: plug in the connector once, and from then on the assistant sees straight into the tool you're working with. No copying, no exports, no “hang on, let me send you that.”

On claude.ai you can connect Gmail, Google Calendar, Google Drive, Notion, Slack, Canva, or social-media ad management, among others — there's a directory with hundreds of other services. Where MCP fits into the bigger picture is covered in the chapter on AI and automation.

A typical scenario

Klára, a marketing manager, is prepping for Monday's meeting. Without connectors, that means: open Notion, find the notes from last week's meeting, copy them, open the ad platform's interface, download this week's numbers, retype them into a spreadsheet, then open the calendar and hunt for when the team has a shared free slot. Sixty to ninety minutes, most of it just jumping between windows.

With connectors in place, she writes three sentences and is done in ten minutes. Her judgment hasn't disappeared — she still decides what actually gets discussed at the meeting. What disappeared is the data-shuttling.

Three mini-scenarios that work right away

• “Find the notes from last meeting in Notion and send me a summary of what's still unfinished.” Claude finds the page itself, goes through the tasks, and returns a list of open items along with who owns each one.
• “Check my calendar and the team's calendar and suggest three one-hour meeting slots for next week, ideally in the morning.” Instead of five emails along the lines of “how about Wednesday?”, you get three real open blocks right away.
• “Put together a performance report on the last ad campaign: compare it to the previous period and write three sentences on what it means.” It pulls the numbers from the ad account itself and writes the comparison and the takeaway.

How to do it

1. Start with one connector, not all of them. In claude.ai's connector settings, turn on the tool you spend the most time in. That's usually mail or a document store.
2. Sign in and review the permissions. The connection runs through your own account, and you can see exactly what the connector is allowed to do. Stick to least privilege: grant only what your scenario needs, and nothing more.
3. Try three read-only requests before you allow writing. For the first week, ask only for things that don't change anything — search, summarize, compare. Once you trust it, add actions that create things.
4. Write requests like you'd talk to a person, not a machine. You don't need to know database or schema names. “Find the latest version of the Smith proposal in Drive and summarize what we promised” is a complete request.
5. Think about the data. The content of your mail and documents flows through the connector. Sensitive material — personal data, finances, health information — belongs strictly on a paid or business account with contractual data protection, not a free chat; anonymize whatever can be anonymized.
6. For companies: build your own server. MCP is an open standard, so developers can write their own MCP server on top of an internal system — inventory, CRM, attendance tracking — and AI works with it just as naturally as it does with Gmail.

The best tools

• The connector directory on claude.ai — ready-made links to hundreds of services; turning one on takes two clicks.
• Gmail and Google Calendar — the fastest payoff; mail and scheduling are where data-shuttling happens most often.
• Notion, Google Drive, and Slack — your organization's memory; useful anywhere the answer exists but nobody can find it.
• A custom MCP server — for internal company systems that don't have a ready-made connector.

What you get out of it

• Time: a conservative 2–4 hours a week for anyone who jumps between mail, documents, and a calendar every day.
• Accuracy: AI works with live data straight from the source, not with whatever you managed to copy.
• Less friction: the “is it even worth copying this over?” calculation disappears — that calculation is what stops a lot of people from using AI at all.
• Control: you grant the permissions and can revoke them at any time. Connecting a tool isn't an irreversible decision.

Pro tip

Don't count on remembering, six months from now, everything your assistant is allowed to do. Put a quarterly “connector review” reminder in your calendar and, when it comes up, turn off anything you haven't used in the last three months. And keep the rule that holds across this whole site: a connector is permission to read and prepare, not to send or pay. Actions with an outward-facing consequence get approved by a human.