AGENTS/skills/meeting-notes/references/teams-transcript-workflow.md

Teams Transcript Processing Workflow

Manual workflow for processing Teams meeting transcripts (.docx) into structured meeting notes with action items.

Table of Contents

1. Workflow Overview
2. Prerequisites
3. Step-by-Step Process
4. Templates
5. Integration Points
6. Best Practices
7. Troubleshooting

---

Workflow Overview

Teams Transcript (.docx)
    ↓
[Manual: Upload transcript]
    ↓
[Extract text content]
    ↓
[AI Analysis: Extract key info]
    ├─→ Attendees
    ├─→ Topics discussed
    ├─→ Decisions made
    └─→ Action items
    ↓
[Create Obsidian meeting note]
    ├─→ Use meeting-notes template
    ├─→ Include transcript summary
    └─→ Extract action items as tasks
    ↓
[Optional: Sync to Basecamp]
    ├─→ Create todos in Basecamp
    └─→ Assign to project

---

Prerequisites

Tools Needed

• Teams (for recording and downloading transcripts)
• Python with python-docx library (for text extraction)
• Obsidian (for storing meeting notes)
• Basecamp MCP (for syncing action items - optional)

Install Dependencies

pip install python-docx

---

Step-by-Step Process

Step 1: Download Teams Transcript

In Teams:

1. Go to meeting recording
2. Click "..." (more options)
3. Select "Open transcript" or "Download transcript"
4. Save as .docx file
5. Note filename (includes date/time)

Filename format: MeetingTitle_YYYY-MM-DD_HHMM.docx

Step 2: Extract Text from DOCX

Python script (/tmp/extract_transcript.py):

#!/usr/bin/env python3
from docx import Document
import sys
import os

def extract_transcript(docx_path):
    """Extract text from Teams transcript DOCX"""
    try:
        doc = Document(docx_path)
        full_text = '\n'.join([para.text for para in doc.paragraphs])
        return full_text
    except Exception as e:
        print(f"Error reading DOCX: {e}")
        return None

if __name__ == "__main__":
    if len(sys.argv) != 2:
        print("Usage: python extract_transcript.py <transcript.docx>")
        sys.exit(1)

    docx_path = sys.argv[1]
    text = extract_transcript(docx_path)
    
    if text:
        print(text)
        # Optionally save to text file
        output_path = docx_path.replace('.docx', '.txt')
        with open(output_path, 'w') as f:
            f.write(text)
        print(f"\nExtracted to: {output_path}")

Usage:

python extract_transcript.py "MeetingName_2026-01-28_1400.docx"

Step 3: AI Analysis of Transcript

Prompt for AI (ask your AI assistant):

Analyze this Teams meeting transcript and extract:

1. Attendees:
   - Names and roles (if mentioned)
   - Who spoke the most
   - Any key stakeholders

2. Topics Discussed:
   - Main topics (3-5 key items)
   - Brief summary of each topic

3. Decisions Made:
   - Clear decisions with reasoning
   - Format: "Decision: [what] - [reasoning]"
   - Include: "Deferred decisions" if applicable

4. Action Items:
   - Extract specific, actionable tasks
   - Include: owner (@mention), due date (if mentioned), priority (implicit from context)
   - Format: "- [ ] Task description #meeting #todo 🔼 👤 @name 📅 YYYY-MM-DD"

5. Next Steps:
   - Follow-up meetings needed
   - Deliverables expected
   - Blockers or dependencies

Format output in markdown ready for Obsidian meeting template.

Step 4: Create Obsidian Meeting Note

Ask AI to:

• Use meeting-notes skill template
• Format extracted content
• Create proper frontmatter
• Add wiki-links to related projects/areas

Template structure (from meeting-notes skill):

---
title: "[Meeting Title]"
platform: teams
date: YYYY-MM-DD
time: HH:mm-HH:mm
duration: "X minutes"
attendees: [names]
transcript_file: "MeetingName_2026-01-28.docx"
project: [[Project Name]]
tags: [meeting, work, teams]
---

## Attendees
- [Name] - [Role]
- [Name] - [Role]

## Topics Discussed
### [Topic 1]
- [Summary]

### [Topic 2]
- [Summary]

## Decisions Made
1. [Decision] - [Reasoning]
2. [Decision] - [Reasoning]

## Action Items
- [ ] [Action description] #meeting #todo 🔼 👤 @name 📅 YYYY-MM-DD
- [ ] [Action description] #meeting #todo 🔽 👤 @self 📅 YYYY-MM-DD

## Next Steps
- [ ] Schedule follow-up meeting
- [ ] Share notes with team

Step 5: Save to Obsidian

Location:

~/CODEX/01-projects/work/[project]/meetings/[topic]-YYYYMMDD.md

Ask AI to:

• Determine project from context
• Create proper folder structure
• Use kebab-case for filename
• Add to project MOC

Step 6: Sync Action Items to Basecamp (Optional)

When to sync:

• Meeting was about a specific project
• Action items have clear owners
• Project uses Basecamp for task tracking

Ask user: "Sync these action items to Basecamp?"

If yes:

1. Delegate to basecamp skill
2. Ask: "Which Basecamp project?"
3. Create todos with:
   • Proper due dates
   • Assignees (from @mentions)
   • Linked to project
4. Confirm: "Created X todos in [project]"

---

Templates

AI Analysis Prompt Template

Copy this prompt for consistent results:

You are a meeting analysis assistant. Analyze this Teams meeting transcript and extract:

1. Attendees:
   - List all participants mentioned
   - Identify speakers (who talked most)
   - Note any key stakeholders (managers, decision-makers)

2. Topics Discussed (3-5 main topics):
   For each topic:
   - Title (2-4 words)
   - Summary (2-3 sentences)
   - Time spent on topic (if discernible from transcript)

3. Decisions Made:
   For each decision:
   - Decision statement (what was decided)
   - Reasoning (brief justification)
   - Consensus level (unanimous / majority / proposed)
   - Format as checklist item: `- [ ] Decision: [text]`

4. Action Items:
   For each action item:
   - Description (specific, actionable verb)
   - Owner (@mention if clear, otherwise "Unassigned")
   - Due date (YYYY-MM-DD if mentioned, else "No deadline")
   - Priority (implicit: ⏫ urgent, 🔼 high, 🔽 low)
   - Format: `- [ ] Task #meeting #todo [priority] 👤 @owner 📅 date`

5. Next Steps:
   - Follow-up meetings needed?
   - Deliverables expected?
   - Blockers or dependencies?

**Output Format**: Markdown ready for Obsidian meeting note template.

**Meeting Type**: [standup / 1:1 / workshop / decision]

Transcript:
[PASTE TEAMS TRANSCRIPT HERE]

Meeting Note Template (Enhanced for Teams)

---
title: "[Meeting Title]"
platform: teams
date: YYYY-MM-DD
time: HH:mm-HH:mm
duration: "X minutes"
attendees: [Name 1 - Role, Name 2 - Role, ...]
transcript_file: "transcripts/[Topic]-YYYYMMDD.docx"
recording_link: "[Teams recording URL if available]"
project: [[Project Name]]
tags: [meeting, work, teams, transcript]
---

## Attendees

| Name | Role | Company |
|-------|-------|---------|
| [Name] | [Role] | [Company] |
| [Name] | [Role] | [Company] |

## Agenda
[If agenda was known in advance]
1. [Item 1]
2. [Item 2]
3. [Item 3]

## Transcript Summary

[AI-generated summary of transcript]

## Topics Discussed

### [Topic 1]
- [Summary points]
- [Time spent: X minutes]

### [Topic 2]
- [Summary points]
- [Time spent: X minutes]

## Decisions Made

1. ✅ [Decision 1]
   - **Reasoning**: [Why this decision]
   - **Owner**: [Who made decision]
   - **Due**: [If applicable]

2. ✅ [Decision 2]
   - **Reasoning**: [Why this decision]
   - **Owner**: [Who made decision]

### Deferred Decisions
- [ ] [Decision deferred] - [Why deferred, revisit date]

## Action Items

- [ ] [Task 1] #meeting #todo 🔼 👤 @owner 📅 YYYY-MM-DD
- [ ] [Task 2] #meeting #todo ⏫ 👤 @owner 📅 YYYY-MM-DD
- [ ] [Task 3] #meeting #todo 🔽 👤 self 📅 YYYY-MM-DD

### Action Item Summary

| Task | Owner | Due | Priority |
|-------|--------|------|----------|
| [Task 1] | @owner | YYYY-MM-DD | ⏫ |
| [Task 2] | @owner | YYYY-MM-DD | 🔼 |
| [Task 3] | @self | N/A | 🔽 |

## Next Steps

- [ ] Schedule follow-up meeting: [Topic] - [Proposed date]
- [ ] Share notes with: [attendee list]
- [ ] Update project status in Basecamp

## Notes

[Additional notes, observations, or clarifications]

## Links

- 📹 Teams Recording: [URL if available]
- 📄 Transcript: [[transcript_filename]]
- 🗄 Project: [[Project Name]]
- 📄 Related Docs: [[Related Outline Doc]](outline://document/abc123)

---

Integration Points

With meeting-notes Skill

Flow:

User: "Process transcript: [file.docx]"

1. Extract text from DOCX
2. Ask AI to analyze transcript
3. AI extracts: attendees, topics, decisions, action items
4. Create meeting note using meeting-notes skill
5. Ask: "Sync action items to Basecamp?"

With basecamp Skill

Flow (optional):

User: "Yes, sync to Basecamp"

1. Ask: "Which Basecamp project?"
2. List available projects
3. For each action item:
   - create_todo(project_id, todolist_id, content, due_on, assignee_ids)
4. Confirm: "Created X todos in [project]"

With obsidian-management Skill

Flow:

1. Create meeting note at: 01-projects/work/[project]/meetings/[topic]-YYYYMMDD.md
2. Update project MOC with link to meeting:
   - Add to "Meetings" section in project _index.md
3. If decision made, create in decisions/ folder
4. If applicable, export decision to Outline wiki

---

Best Practices

During Meeting

1. Use Teams recording: Get transcript automatically
2. Name attendees: Add their roles to transcript
3. Speak clearly: Improves transcript accuracy
4. Agenda first: Helps AI structure analysis

Processing Transcripts

1. Process quickly: Within 24 hours while fresh
2. Clean up text: Remove filler words (um, ah, like)
3. Be specific: Action items must be actionable, not vague
4. Assign owners: Every action item needs @mention
5. Set due dates: Even if approximate (next week, by next meeting)

Storage

1. Consistent location: All work meetings in project/meetings/
2. Link everything: Link to project, related docs, areas
3. Tag properly: #meeting, #work, #teams, #transcript
4. Archive old: Move completed project meetings to archive/

Basecamp Sync

1. Sync important meetings: Not every meeting needs sync
2. Use project context: Sync to relevant project
3. Verify in Basecamp: Check todos were created correctly
4. Follow up: Check completion status regularly

---

Troubleshooting

Transcript Won't Open

Problem: DOCX file corrupted or wrong format

Solution:

1. Re-download from Teams
2. Try opening in Word first to verify
3. Use alternative: Copy-paste text manually

AI Misses Action Items

Problem: Transcript analysis misses clear action items

Solution:

1. Manually add missed items to meeting note
2. Reprompt AI with specific context: "Review transcript again, focus on action items"
3. Check transcript: Was the audio clear?

Wrong Project Assigned

Problem: Meeting note created in wrong project folder

Solution:

1. Move file to correct location
2. Update links in project MOCs
3. Use consistent naming conventions

Basecamp Sync Fails

Problem: Todos not created in Basecamp

Solution:

1. Check Basecamp MCP is connected
2. Verify project ID is correct
3. Check assignee IDs are valid
4. Check todo list exists in project
5. Retry with fewer items

---

Example End-to-End

Input

Teams transcript: api-design-review_2026-01-28_1400.docx

AI Output

## Attendees
- Alice (Product Owner)
- Bob (Lead Developer)
- Charlie (Tech Lead)
- Sarah (UX Designer)

## Topics Discussed
1. API Authentication Design (20 min)
2. Rate Limiting Strategy (15 min)
3. Error Handling (10 min)

## Decisions Made
1. Use OAuth2 with refresh tokens - Industry standard, better security
2. Implement 1000 req/min rate limit - Based on load tests

## Action Items
- [ ] Create OAuth2 implementation guide #meeting #todo 🔼 👤 @alice 📅 2026-02-05
- [ ] Document rate limiting policy #meeting #todo 🔼 👤 @bob 📅 2026-02-10
- [ ] Update error handling documentation #meeting #todo 🔽 👤 @sarah 📅 2026-02-15

Obsidian Note Created

File: ~/CODEX/01-projects/work/api-integration-platform/meetings/api-design-review-20260128.md

Basecamp Sync

Project: API Integration Platform Todos created: 3

• OAuth2 guide (assigned to Alice, due 2026-02-05)
• Rate limiting (assigned to Bob, due 2026-02-10)
• Error handling (assigned to Sarah, due 2026-02-15)

---

Automation (Future n8n Implementation)

When n8n is added, automate:

1. Watch transcript folder: Auto-trigger on new .docx files
2. AI auto-analysis: Use AI API to extract meeting info
3. Auto-create meeting notes: Save to Obsidian automatically
4. Auto-sync to Basecamp: Create todos for action items
5. Send notifications: "Meeting processed, X action items created"

Workflow diagram:

Teams transcript folder
    ↓ (n8n: Watch folder)
[Trigger: new .docx]
    ↓
[Extract text]
    ↓
[AI Analysis]
    ↓
[Create meeting note]
    ↓
[Sync to Basecamp] (conditional)
    ↓
[Send ntfy notification]

---

Quick Reference

Action	Tool/Script
Download transcript	Teams UI
Extract text	python extract_transcript.py
Analyze transcript	AI assistant prompt
Create meeting note	meeting-notes skill
Sync to Basecamp	basecamp skill
Store in Obsidian	obsidian-management skill
Export decision	outline skill (optional)