Connect Meeting Notes to Claude & ChatGPT With the Granola MCP Connector
Kevin Stratvert walks through using the Granola MCP connector to pipe meeting transcripts into Claude, ChatGPT, and Claude Cowork. Once connected, the AI pulls meeting context automatically — without being asked — to generate proposals, risk assessments, action-item spreadsheets, draft emails, and calendar events from what was discussed.
"Your Meetings in Claude & ChatGPT: Granola Connectors" by Kevin Stratvert — Watch on YouTube →
Key Takeaways
- Granola transcribes any meeting, then exposes it over MCP. It captures Zoom, Google Meet, and Microsoft Teams calls, then makes those transcripts available to any MCP-compatible AI — Claude, ChatGPT, or another LLM.
- Setup in Claude is a few clicks. Customize → Connectors → Browse → search "Granola" → Add → sign in. You then set per-tool permissions: always allow, or ask each time.
- Claude invokes the connector on its own. Stratvert never told it to check Granola; given a proposal task, Claude recognized it should pull the relevant meeting and did so automatically.
- Stacking connectors builds a knowledge base. Pair Granola with Google Drive / Gmail / Calendar (or Microsoft 365 + Slack + Asana / monday.com) so the assistant draws on meetings, docs, email, and calendar together.
- Hand the same connectors to Claude Cowork for multi-output projects. In one run it produced a proposal, a risk-assessment doc, draft intro emails per department head, and calendar milestones — all grounded in the meeting notes.
Connectors that read your meetings, email, and drive grant the model broad access to private data. Review each connector's permission scope and prefer "ask each time" for anything sensitive — see our security center and MCP explainer.
Compare assistants for this kind of work: ChatGPT hub · Claude Cowork hub · decision guide.
What you can actually set up from this
Extracted from the video's own transcript — the specifics the original summary left out.
Reproducible steps
- Add the Granola connector in Claude
Go to Customize, then Connectors, browse or search for Granola, add it and sign in. You can then set each tool to always allowed or to ask permission each time.
- Use it from a normal chat
Attach your earlier proposals and a spreadsheet of costs, then ask Claude to refer to a specific meeting and write a proposal consistent with the attachments. Claude decides on its own to call the Granola connector, and returns a downloadable Word document.
- Combine connectors for a knowledge base
Add Google Drive, Gmail and Calendar, or Microsoft 365 (which includes Teams), Slack, and project tools such as monday.com or Asana. One prompt can then pull action items from a meeting and create a sheet in Drive with a task-assignment column.
- Hand multi-step work to Cowork
In Claude Cowork the same connectors are available: one request produced a proposal, a risk assessment, draft emails to each department head in Gmail and milestone entries in Google Calendar.
Gotchas
- This is a partnership video with Granola; it ends with a promotional code, which isn't reproduced here.
- MCP is a standard, so the same connector works with ChatGPT and other MCP clients; the video shows only Claude.
More ChatGPT news
The reported OpenAI agent breach of Hugging Face, and five rules for running your own agents: sandbox, deny network, scoped credentials, no shared writable folders, verify outside
Hy-Memory: Tencent's open 30B-A3B memory writer, served locally with vLLM and tested on a Seattle-to-Sydney fact update
Getting started with the Codex desktop app: where it runs, which model and effort to pick, /goal, recording a skill, plugins and your dot
OpenAI Dots hands-on: setup, the privacy toggle to flip first, what it can do and where it was buggy
Build your own agentic harness in Python: the loop, tool calling, MCP servers, memory and model switching
GPT-6 Astra Ultrafast in Codex: how to switch it on, what it costs, what it built