FlowPocket
Tools › Audio tools › Meeting Minutes Builder

🗒️Meeting Minutes Builder

Turn a meeting recording into a minutes template with fields for title, date, attendees, summary, decisions and to-dos. Transcription is done by an on-device AI; the summary is written by you, not AI.

Your audio is never sent to a server (the AI model itself is downloaded from a third-party site the first time you use this tool)

Meeting details (you can also edit these in the text later)
Provide audio (choose a file, or record live)

Transcription settings

Minutes (fully editable here)

Summary, Decisions and To-do are left blank rather than filled in by AI — read the transcript and add them yourself. The first run downloads the AI model (fast after that).

How to use

  1. Enter the meeting name, date, attendees and agenda (optional — you can proceed without them and edit the text afterward).
  2. Choose an audio file, or record the meeting live with "Record with microphone".
  3. Pick accuracy and language, then press "Transcribe and build minutes".
  4. Read the resulting draft and fill in Summary, Decisions and To-do to finish it off.

This tool fills in a template — it doesn't summarize for you

This tool transcribes your meeting audio with an on-device AI (Whisper), breaks it into paragraphs based on pauses, and pours the result into a template with meeting name, date, attendees, summary, decisions and to-dos. It does not, however, write the contents of "Summary" or "Decisions" for you — this tool is meant to be finished by reading the transcript and adding the key points in your own words.

There are two reasons for this design. First, automatic AI summarization can misjudge what matters and miss real decisions. Second, since the audio never leaves your browser and no meeting audio is sent to a server, running an AI capable of quality summarization on-device isn't realistic today. Transcription gets AI help; deciding what matters stays with a human.

How the paragraph breaks work

The transcript automatically breaks into a new paragraph wherever there's a pause of roughly 2 seconds or more between chunks of speech. Fast back-and-forth conversation tends to stay in one paragraph, while natural pauses split things up. If a break doesn't line up with an actual topic change, just edit the text directly — the output box is fully editable.

FAQ

Is my audio uploaded anywhere?
No. Transcription and formatting all run inside your browser. Only the AI model itself is downloaded from a distribution site on first use — neither your audio nor the minutes content is ever sent anywhere.
Will it pull out decisions and to-dos automatically?
No. This tool gets you to the point of a formatted template with the transcript inside — the Decisions and To-do contents are meant to be added by you after reading the body. That's a deliberate choice to keep interpretation in human hands.
I just want a timestamped transcript, not a minutes template.
If you don't need the minutes format and just want subtitles (SRT/VTT) or a timestamped transcript, use the Audio Transcription tool instead.
Does it work for long meetings?
Yes, but since the AI runs in your browser, long audio takes longer to process. As a guide, audio over an hour is best run on a PC.
Will the browser show a "page not responding" warning during processing?
No. The heavy transcription work runs in a background thread (a Web Worker) separate from the page itself, so the display keeps updating throughout, and the browser never prompts you to wait or exit.

Related tools