Granola Meeting Extraction

Extract and archive Granola meeting notes with optional AI intelligence.

When to Use

• User asks: "Extract my last 7 days of Granola meetings"
• User asks: "Analyze my meeting with [Person]"
• User asks: "What companies did I discuss this week?"
• User asks: "What TODOs came from my meetings?"

Requirements

1. Granola installed on macOS with at least one recorded meeting
2. OpenAI API key stored in ~/.granola-archivist/.env (optional, only for AI extraction)

Setup Instructions

First-Time Setup

Tell the user:

To enable AI extraction, you need to set up your OpenAI API key.

Run these commands:
mkdir -p ~/.granola-archivist
echo "OPENAI_API_KEY=sk-your_key_here" > ~/.granola-archivist/.env

Then try: "Extract my last 7 days of Granola meetings"

Checking Setup

Before running AI extraction, check if .env exists:

• File exists at ~/.granola-archivist/.env → proceed
• File missing → show setup instructions above

Usage

Extract Recent Meetings

User: "Extract my last 5 Granola meetings"

You should:

1. If AI is requested, check setup (~/.granola-archivist/.env exists)
2. Run: python skills/granola/extract.py --days 7 (add --no-intelligence if no AI)
3. Show results: Meeting titles, attendees, where files were saved

Get Intelligence from Meetings

User: "What companies did I discuss this week?"

You should:

1. Extract last 7 days of meetings
2. Parse output files from ~/.granola-archivist/output/
3. Summarize companies mentioned across all meetings

Extract TODOs

User: "What action items came from my meetings?"

You should:

1. Extract recent meetings with intelligence (requires API key)
2. Parse TODO sections from output files
3. Present consolidated TODO list with owners and deadlines

Output Format

Files are saved to: ~/.granola-archivist/output/

Filename format: YYYY-MM-DD-meeting-title-attendee-name.md

Each file contains:

• Meeting metadata (title, date, attendees, location)
• Your manual notes from Granola
• AI-generated summary from Granola
• Extracted intelligence:
  • People (with roles, insights, network connections)
  • Companies discussed (with stage, funding, context)
  • Network connections (who knows who)
  • TODOs (with owners, deadlines, context)
  • Key insights (investment opportunities, market insights)

Examples

Example 1: Extract Recent Meetings

User: "Extract my last 3 Granola meetings"

You: "I'll extract your last 3 Granola meetings. Let me check your setup first..."

[Checks ~/.granola-archivist/.env exists]

You: "Setup looks good. Extracting meetings..."

[Runs: python skills/granola/extract.py --days 7 --limit 3]

You: "✓ Extracted 3 meetings:
- 2026-01-14: Fund Discussion with Srihita Vatsavaya
- 2026-01-13: Rollout AI Sync with Camden Ko
- 2026-01-12: Notion Strategy with Jade Doan

Files saved to: ~/.granola-archivist/output/

Would you like me to:
1. Show summaries of all 3 meetings
2. Extract companies discussed
3. Pull out action items"

Example 2: Missing Setup

User: "Extract my Granola meetings"

You: "If you'd like AI extraction, set up your OpenAI API key first. Please run:

mkdir -p ~/.granola-archivist
echo 'OPENAI_API_KEY=sk-your_key_here' > ~/.granola-archivist/.env

Replace 'sk-your_key_here' with your actual OpenAI API key.

If you prefer no AI, I can still export your meetings without a key."

Example 3: Extract Companies

User: "What companies did I discuss this week?"

You: "Let me extract your meetings from the last 7 days and analyze them..."

[Extracts meetings, parses output files]

You: "Here are the companies discussed across 5 meetings this week:

**Rollout AI** (mentioned 8 times)
- Stage: Series A
- Discussed in: Camden meeting, Investor call
- Context: Potential investment target, strong growth

**Notion** (mentioned 5 times)
- Stage: Late-stage
- Discussed in: Jade meeting
- Context: Partnership opportunity

**OpenAI** (mentioned 3 times)
- Discussed in: Multiple meetings
- Context: API integration plans

Would you like more details on any of these?"

Cost Information

Using GPT-5-nano (cheapest model):

• Cost per meeting: ~$0.0009
• 30 meetings/month: ~$0.03

Troubleshooting

"Granola cache not found"

• Ensure Granola is installed
• Record at least one meeting
• Check: ~/Library/Application Support/Granola/cache-v3.json exists

"OpenAI API key not found"

• Setup .env file: ~/.granola-archivist/.env
• Format: OPENAI_API_KEY=sk-...

Legacy config

• If ~/.granola-claude/ exists, it will still be detected

"No meetings found"

• Check date range (default: last 7 days)
• Ensure Granola has meetings in that period

Privacy

• No data collection or tracking
• Transcripts stay on your machine
• Only sent to your OpenAI account (you control your data)
• Your API key = you control your data