From meeting to meeting notes in four steps

There's no magic gadget behind good meeting notes — just a simple workflow: record the meeting, transcribe it with speaker labels, and let an AI write the notes. Here's the exact process, and why the speaker-label step is the part most tools skip.

1Record the meeting

Any recorder works: your phone's built-in app (Voice Memos on iPhone, Recorder on Android), a third-party app like VoiceNotes, or dedicated hardware — a handheld recorder, or wireless podcast-style mics like the DJI or Rode kits you see on YouTube channels. Put the phone or mic in the middle of the table, hit record, and forget about it. Two tips that make everything downstream better:

1. Get consent. Tell the room you're recording for note-taking purposes. It's polite, and in many states it's the law.
2. Minimize crosstalk where you can. Speaker labeling works best when people mostly take turns; a recording where everyone talks over each other is harder for any tool.

2Get the file to your computer

However is easiest for your setup: AirDrop from an iPhone to a Mac, Quick Share or Google Drive from Android, email it to yourself, sync through Dropbox/OneDrive/iCloud, or plug the recorder in with a cable and copy the file. Any common audio format is fine (m4a, mp3, wav).

3Transcribe it — with speaker labels

Drop the file into LymeScribe's Transcribe files card and choose the output with speaker labels. A few minutes later you have a transcript where every line says who said it.

The labels come out as Speaker 1, Speaker 2, and so on — no software can know a stranger's name from their voice. That's why LymeScribe includes a Name speakers step: it plays you a short clip of each voice, you type the name, and the whole transcript updates to use real names. Thirty seconds of your input, and every line now reads Phil: instead of Speaker 2:.

This is the step that separates this workflow from "just use my phone's transcription." Here's the same moment of a meeting, both ways:

...so we should push the closing to
March. I don't think the lender will
go for that. They said thirty days.
That was before the appraisal came
back. Okay so who's calling them?
I can do it Monday.

Mike: ...so we should push the
closing to March.
Phil: I don't think the lender will
go for that. They said thirty days.
Mike: That was before the appraisal
came back.
Susan: Okay so who's calling them?
Phil: I can do it Monday.

Read the left version and try to write down who committed to what. You can't — and neither can an AI. The right version makes decisions, disagreements, and action items unambiguous. When the AI writes "Phil will call the lender Monday," it's because the transcript actually says so.

4Turn the transcript into meeting notes

Paste the transcript into Claude, ChatGPT, or whichever AI you use, with a prompt like:

Here is a transcript of a meeting with speaker labels. Write meeting
notes with: a one-paragraph summary, decisions made, action items
(with who owns each one and any deadlines), and open questions.

Because you named the speakers in step 3, the transcript already carries real names — so the notes come out reading like a person took them, with action items assigned to actual people. From the same transcript you can also ask for a client-ready recap email, a task list, or a follow-up agenda.

Where this is headed

Today this is a four-step process. We're working on collapsing it — tighter hand-offs from phone to transcript to notes — but each step is honest work done well, and the whole thing takes about five minutes of your attention per meeting. The transcription itself runs locally on your machine: nothing is uploaded anywhere unless you choose to paste it into an AI.