Tools, connectors & MCP — Agents & Automation
Models are brains without hands
On its own, a model can only produce text. Tool use (function calling) gives it hands: the app describes available tools — "search the web", "read a file", "create a calendar event" — and the model asks to use them. The app runs the tool and hands the result back.
Every "connect Google Drive / Slack / GitHub" button you see is tool use underneath.
📝 Read the full lesson notes
Models are brains without hands
On its own, a model can only produce text. Tool use (function calling) gives it hands: the app describes available tools — "search the web", "read a file", "create a calendar event" — and the model asks to use them. The app runs the tool and hands the result back.
Every "connect Google Drive / Slack / GitHub" button you see is tool use underneath.
MCP: the USB-C port for AI
Before MCP, every AI app needed a custom integration for every tool. The Model Context Protocol — introduced by Anthropic in 2024 and now supported across the industry — is one standard plug: build an MCP server for your tool once, and any MCP-compatible assistant or agent can use it.
There are MCP servers for GitHub, Slack, Google Drive, Notion, databases, browsers, and thousands more.
Connecting tools wisely
- Start read-only (search, read) before granting write access (send, edit, delete).
- Connect only what the task needs — fewer tools, fewer mistakes.
- Prefer official connectors and well-known MCP servers; a connector can see whatever you give it.
- Review what an agent did with a tool, especially the first few times.
Try it: a connected question
Search my Google Drive for anything about [project]. Summarize the current status, open questions, and who owns what — with links to the source docs.
Part of 🤖 Agents & Automation in the free AI Bootcamp.