$cd ../use-cases/
β‘ ProductivityHigh Value15 min setup
$ cat meeting-notes-generator.md
await transcribeMeeting(audio, { actionItems: true })
/** Record a call β get structured notes with action items sent to Slack in 60 seconds. */
Lisa manages a 20-person remote team across 3 time zones. Every week, she sat through 8 meetings and spent 3+ hours afterwards writing up notes and action items. Half the team never read them. After deploying OpenClaw with Whisper, her post-meeting workflow became: drop the recording file into a shared folder β OpenClaw transcribes, extracts decisions and action items with owners and deadlines β sends the structured note to Slack and assigns tasks in Linear. Total time: 60 seconds. Lisa estimates she saved 12 hours per month, and action item completion rates jumped from 40% to 85% because assignments were clear and immediate.
how_it_works.md
βοΈ How It Works
1.
Record your meeting (Zoom/Meet/Phone)
Use any recorder. Drop the audio file into the OpenClaw watch folder.
2.
Whisper transcribes the audio
Runs locally. No audio ever sent to a cloud API.
3.
LLM structures the transcript
Extracts: summary, decisions, action items with owners and deadlines.
4.
Sends the note to Slack/Notion/Email
Wherever your team works. Includes a searchable archive.
example_note.md
π Meeting Note Example
# Q1 Planning Call β 2026-03-03
## Summary
Team aligned on product roadmap Q1. Prioritized mobile app and API expansion. Budget approved.
## Action Items
β’ @alice: Draft API spec by March 10
β’ @bob: Set up mobile dev environment
β’ @carol: Schedule design review
β’ @bob: Set up mobile dev environment
β’ @carol: Schedule design review
## Decisions
β’ Use React Native for mobile
β’ Launch beta April 1
β’ Launch beta April 1
β FAQ
Q1. Does it work with Zoom/Google Meet/Teams?
Yes. OpenClaw processes any audio file (MP3, WAV, M4A). Record via Zoom's built-in recorder, Google Meet recording, or any third-party tool, then drop the file into OpenClaw's watch folder.
Q2. Is my meeting data private?
Completely. Whisper transcription runs locally on your hardware. The audio file never leaves your server. No cloud APIs are involved in the transcription process.
Q3. How accurate is the transcription?
Whisper large-v3 achieves 95%+ accuracy for English and 90%+ for most languages. For technical jargon, you can add custom vocabulary lists.
Q4. Can it handle multiple speakers?
Yes. OpenClaw uses speaker diarization to identify different speakers in the meeting. Action items are attributed to the correct person.
Q5. What format are the notes in?
Structured markdown with sections: Summary, Key Decisions, Action Items (with owner + deadline), and Discussion Points. Exported to Slack, Notion, email, or any webhook.