If you've ever used Claude to complete tasks after a meeting, you know the routine: find the transcript, copy a chunk of it, paste it into the conversation, add your prompt, and finally get to the actual work. Every single time.
It works, but the copy-pasting is the problem. It's friction before the thinking even starts, and it means you're only ever working with whatever you remembered to grab.
An MCP connection removes that step. Once it's set up, Claude can read your meeting history directly. No copying. No pasting. Just ask, and the context is always there.
This post covers how to connect the MCP and six ways to start using it today.
Setting up Supernormal MCP in Claude
If you're using Claude on the web (Pro, Max, Team, or Enterprise), go to claude.ai/customize/connectors, add a custom connector, name it Supernormal, and paste in the server URL. Authenticate, and you're done.
If you're using Claude Code, run one command:
Supernormal MCP also works with Cursor, ChatGPT, and any other MCP-compatible tool. Full setup instructions for each are in the Help Center.
Once the MCP is connected, give it a quick test in Claude for the web or Claude Code by asking something like:
If Claude comes back with real details from your transcript, you're ready to go.
6 ways to use your meeting notes in Claude
1. Turn any meeting into a to-do list
Every meeting ends the same way: action items get assigned, people head back to their desks, and those items exist only in a transcript nobody is going to re-read in full. Someone will follow up on the obvious ones. The rest will quietly fall through.
After a planning session or standup, the last thing you want to do is paste a transcript into Claude and ask it to find the action items. You shouldn't have to. Claude can go get them directly.
Try this:
Because Claude is reading the full transcript, it catches the follow-ups that got mentioned in passing, not just the ones that came with a name attached. The offhand "can you look into that?" gets captured the same as the formally assigned task. You get a complete list, not a partial one.
From there, you can ask Claude to format it as a checklist, draft it into a follow-up email, or push it into whatever system your team uses to track work. Note that you can also run all these prompts in Supernormal itself after your meeting, but if you're working on other tasks it tools like Claude, it can be useful to have everything in one place.
2. Find what clients actually asked for
When you're about to commit to a project, there's usually a gap between what the team thinks clients want and what clients actually said. Closing that gap normally means going back through call recordings, and nobody has time to do that thoroughly.
The feedback you need is spread across many conversations. One clients mentioned it in passing three weeks ago. Another brought it up as a blocker. A third had a workaround. None of that makes it into a summary doc. It just sits in transcripts.
Claude searches across all the meetings in a Supernormal project, and any other project context you have stored in a Claude project.
Try this:
You get the full picture across every relevant call and Claude output, not just the ones you happened to remember, and not filtered through whoever wrote up the notes. That's the difference between working from gut feel and working from real context.
3. Organize your meetings into projects automatically
Your meeting library grows fast. Discovery calls, client check-ins, internal planning sessions, stakeholder updates: they all pile up in one long list. Getting them organized into coherent projects means going through each one manually, figuring out which belong together, and naming the groupings in a way that actually makes sense.
It's tedious work. You know it needs to happen, but you also know it will take an hour you don't have, so it sits there unorganized.
Claude can do it in seconds.
Try this:
Claude reads across your meeting history, identifies the patterns, groups related conversations, and suggests project names that reflect what the work is actually about. Not generic labels like "Client Meetings" but specific ones like "Q2 Brand Refresh" or "Onboarding Flow Redesign."
From there, you can ask Claude to create those projects in Supernormal and move the meetings into them. Or if you want to refine the structure first, ask it to suggest a different breakdown. The point is you're reviewing and approving, not starting from scratch and guessing which meetings go where.
This is especially useful when you're new to a team and inheriting a backlog of unstructured meeting notes. Instead of trying to piece together what each conversation was about, Claude does the first pass for you.
4. Turn a design meeting into a design brief
A good design kickoff or client discovery session produces a lot of raw material: goals, constraints, references, feedback, open questions. Then the meeting ends and someone has to shape all of that into a brief the team can actually work from.
That part is slow. Going back through the recording, finding the key moments, pulling them into a doc, and making it coherent is an hour of work that feels like it shouldn't exist.
Skip it. Ask Claude to write the brief directly from the meeting.
Try this:
The output won't be final and you'll edit it, but you're editing instead of drafting from scratch. More importantly, you're starting from what was actually said, not from your memory of it, which is already getting fuzzy.
If you want a head start on structure, we have a design brief template you can use as the basis for your prompt. You can also pull from multiple meetings: if the brief needs client requirements from one call and internal constraints from a separate team session, ask Claude to combine both. Then you can ask Claude to push the brief into whichever collaboration tools you use and that you have connected to Claude, like Notion, Google Slides, or others.
5. Build onboarding docs from past discussions
When a new person joins and documentation is thin, the knowledge they need isn't missing. It's scattered across past conversations. How a process works, why a decision was made a certain way, what was tried before: it all exists somewhere in your meeting history.
Normally getting to it means tracking down relevant meetings one by one and hoping you got them all. Claude reads across all of them at once.
Try this:
This works for onboarding, but it's just as useful any time you need to explain something that's never been written down. Instead of finding a quiet hour to write a doc from scratch, you ask Claude to pull together what your team has already discussed.
6. Clear your post-call admin in two minutes
Client-facing teams know what comes after every call: write the follow-up email, update the project management tool, log the next steps, assign the action items. Done carefully, it takes 20 to 30 minutes. Done quickly, things get missed.
A customer success manager shared this workflow on Reddit recently: after every call, they drop the transcript into Claude, which drafts a follow-up email straight to their drafts folder and generates structured CRM notes with a summary, next steps, and action items. The whole thing takes two minutes. They review and approve everything before anything goes out. Nothing reaches a customer without them seeing it first.
With Supernormal MCP, the transcript step disappears. Claude already has it.
Try this:
One thing worth highlighting from that Reddit thread: the consistency gain matters as much as the time saving. When you write notes manually, the detail depends on how tired you are. When Claude does it, every record looks the same. Six months later, when you're reviewing a deal or handing off an account, that adds up.
Prefer to stay in Supernormal?
Not everyone wants to add another tool. If that's you, everything above is also available directly inside the Supernormal web app. Ask questions, generate briefs, pull action items, without needing a separate AI subscription.
The MCP connection is for teams who are already in Claude, Cursor, or another AI tool and want their meeting context to follow them there.

