Meeting Recording to Summary

The Problem

You record every client call. Every team sync. Every project kickoff. The recordings pile up in Google Drive. Good intentions, zero follow-through.

Three weeks later, someone asks "what did we decide about the logo?" and you're scrubbing through a 47-minute video looking for a 30-second conversation. Or worse, nobody remembers, and you make the same decision again.

A marketing agency in SLO told me they had 200+ meeting recordings in their Drive. Total watch time: maybe 3 of them. Hundreds of hours of context, insights, and decisions—completely inaccessible.

The meetings happened. The knowledge died.

What This Does

• Watches your recordings folder in Google Drive for new audio or video files
• Transcribes with Whisper—OpenAI's speech-to-text model, highly accurate even with accents and crosstalk
• Generates a structured summary with executive overview, key discussion points, and overall meeting vibe
• Extracts action items with who's responsible and deadlines (when mentioned)
• Lists decisions made so you have a record of what was agreed
• Posts to Slack with the summary and transcript in a thread
• Optionally emails the team for those who prefer inbox over Slack
• Logs everything to a spreadsheet for searchable meeting history

Client calls used to disappear into the void. Now everyone gets a summary within an hour of the call ending—even people who weren't there.

How It Works

Here's the flow:

1. Recording lands in folder—you can set Zoom, Google Meet, or any tool to auto-save there
2. n8n detects the new file and filters for audio/video formats
3. Whisper transcribes the entire recording to text (handles accents, multiple speakers, technical terms)
4. GPT-4o-mini analyzes the transcript and extracts: title, summary, key points, decisions, action items, and meeting sentiment
5. Summary posts to Slack in your #meeting-summaries channel
6. Full transcript replies as thread for reference
7. Email goes out (optional) to specified recipients
8. Everything logs to Google Sheets for searchable history

A 60-minute meeting becomes a 2-minute read. Action items are clear. Decisions are documented. Nobody asks "what did we say about X?" anymore—they just search the sheet.

The Summary Format

Every meeting summary includes:

• Meeting title—AI infers from context if not stated
• Duration estimate—rough length based on transcript
• Attendees—names extracted when mentioned
• Sentiment—productive, challenging, routine, or urgent
• Executive summary—2-3 paragraph overview
• Key discussion points—bullet list of main topics
• Decisions made—what was agreed during the meeting
• Action items—task, owner, and deadline in a table
• Follow-up notes—next meeting or pending items

The AI is trained to be thorough but concise. It won't give you a transcript rehash—it gives you what you need to know without reading 50 pages.

What You'll Need

• n8n instance—Self-hosted (free) or n8n Cloud ($20/month)
• Google Drive—A folder for meeting recordings
• OpenAI API key—For both Whisper (transcription) and GPT (summarization)
• Slack workspace—Create a #meeting-summaries channel
• Gmail (optional)—If you want email summaries too
• Google Sheets—For logging and searchable history

Cost estimate: Whisper costs about $0.006/minute of audio. GPT-4o-mini costs roughly $0.001 per summary. A 60-minute meeting costs about $0.40 to process. Do 50 meetings/month for ~$20 in API costs.

Your Spreadsheet Setup

Create a new Google Sheet called "Meeting Log" with these columns:

• Date—When the meeting was processed
• Title—AI-generated meeting title
• Duration—Estimated length
• Attendees—Who was there
• Summary—Executive overview
• Key Points—Bullet list of discussion items
• Decisions—What was agreed
• Action Items—Task/owner/deadline
• Sentiment—Meeting vibe
• Recording Link—Link back to original in Drive

Now you have a searchable database of every meeting. "What did we decide about pricing?" Ctrl+F → found in March 12th client call.

Setup Instructions

1. Create your recordings folder in Google Drive (e.g., "Meeting Recordings")
2. Configure your recording tool to save to this folder (Zoom, Meet, Loom all have auto-upload options)
3. Create the Slack channel #meeting-summaries
4. Create the spreadsheet with columns listed above
5. Import the workflow—Download the JSON and import into n8n
6. Connect Google Drive—Add credentials, update the folder ID to your recordings folder
7. Connect OpenAI—Add your API key (used for both Whisper and GPT)
8. Connect Slack—Add credentials, update channel name if different
9. Connect Gmail (optional)—Add credentials if you want email summaries
10. Connect Google Sheets—Link your Meeting Log spreadsheet
11. Test with a short recording—Drop a 2-3 minute test audio in the folder
12. Activate—Turn on the workflow and start capturing meeting knowledge

Tips for Better Results

Whisper is remarkably good, but you can help it:

• Use decent audio quality—Built-in laptop mics are fine, but external mics are better
• Minimize background noise—Coffee shop recordings are harder to transcribe
• One speaker at a time—Cross-talk confuses any transcription AI
• Say names explicitly—"Good point, Sarah" helps the AI know who said what
• State action items clearly—"So the action item is: John will send the report by Friday"

The AI will do its best regardless, but clear audio and explicit statements give you better summaries.

Customization Ideas

Once the basics work:

• Different channels for different meeting types—Client calls to #client-calls, internal syncs to #team-syncs
• Action item routing—If someone's assigned a task, DM them directly
• Weekly digest—Summarize all meetings from the week into one Monday email
• Client portal integration—Auto-share relevant meeting summaries with clients
• CRM logging—Push client meeting summaries to their HubSpot/Salesforce record
• Project management—Create tasks in Asana/Trello from extracted action items
• Searchable archive—Use a tool like Notion for richer meeting notes storage

A Simpler Alternative

If you just want transcripts without the full summary workflow:

1. Upload recordings manually to Whisper.ai or use OpenAI's web interface
2. Copy transcript into ChatGPT and ask for a summary
3. Paste summary into Slack yourself

Takes about 5-10 minutes per meeting instead of automatic. Good for low volume or testing whether this is valuable before automating.

Real Impact

A web agency I work with records every client meeting. Before this automation, those recordings sat in Drive collecting dust. 90% never got watched. Decisions got forgotten. Same questions came up meeting after meeting.

Now every project channel gets a summary within an hour. The project manager doesn't have to take notes during calls—she can actually listen and engage. New team members can get context on a client by reading 5 summaries instead of watching 10 hours of video.

The best part? Six months of meeting knowledge became searchable. "What did they say about the rebrand budget?" Used to be a 30-minute video hunt. Now it's a Ctrl+F in a spreadsheet.

Meetings still happen. But the knowledge doesn't die anymore.