Skip to main content

​
Meeting Intelligence Pipeline

​
Overview

Your project meetings contain critical decisions, action items, and strategic discussions - but most of that context gets lost in scattered notes or forgotten conversations. Meeting Intelligence Pipeline automatically processes every recorded meeting, extracts structured minutes and action items, and files everything to your project vault where SupaSherpa can reference it. No more manual note-taking. No more lost commitments. Just meetings that ship outcomes.

​
Step-by-Step Guide

​
Automatic Processing (tl;dv Integration)

  1. Record your meeting - Use tl;dv to record any Zoom, Google Meet, or Teams call with your team, sherpa, or stakeholders.
  2. Processing happens automatically - When the meeting ends, tl;dv sends the recording and transcript to Fiftyknots. The platform queues the meeting for processing (typically completes within 2 minutes for a 60-minute meeting).
  3. Receive structured minutes - Once processing completes, you receive:
    • Meeting summary with 3-5 key discussion points
    • List of decisions made
    • Extracted action items with owners and deadlines
    • Full transcript for reference
  4. Action items delivered to owners - Each person who received an action item gets notified via their preferred channel (Slack, WhatsApp, email, or in-platform notification). The system respects quiet hours (no notifications between 22:00-07:00 local time).
  5. Minutes filed to your vault - Meeting notes automatically appear in your Data Room under the “Meeting Notes” section. The AI can now reference this meeting in future conversations.

​
Manual Meeting Creation

If you have meeting notes but didn’t record via tl;dv:
  1. Navigate to your project dashboard - Visit the relevant job in your pipeline.
  2. Create manual meeting entry - Use the API endpoint or future UI to submit: meeting title, date, duration, attendees, and raw notes or transcript text.
  3. Processing follows same pipeline - The system processes your notes identically to recorded meetings, extracting action items and filing to the vault.

​
Searching Meeting History

  1. Access meeting search - Use the meeting history API (future UI in Data Room or project dashboard).
  2. Search by keyword or date range - Find specific discussions, decisions, or commitments from past meetings.
  3. Review meeting details - Click any result to see full minutes, action items, and original transcript.

​
Common Questions

Q: What happens if the meeting includes people not registered on Fiftyknots?
A: The system extracts action items for everyone mentioned in the meeting. For unmatched people (not in the platform), the action item is flagged and the meeting organizer receives a notification to manually assign it. Platform users receive their items automatically. Q: Can SupaSherpa reference what was discussed in my meetings?
A: Yes. Once a meeting is processed and filed to your vault, SupaSherpa can retrieve and reference specific decisions, action items, and discussion points when you ask about recent meetings or project context. If a meeting is still processing, SupaSherpa will tell you it’s not yet available rather than saying “I don’t know.” Q: What if I receive a duplicate action item from multiple meetings?
A: Each meeting creates its own action items based on what was discussed. The system doesn’t automatically deduplicate across meetings, but you can see the source meeting for each item. If you complete an item in one context, mark it complete to remove it from your active list. Q: How do I know if my action items are overdue?
A: Action items with deadlines automatically track status. SupaSherpa will proactively mention overdue items when you chat 3+ days after the deadline. You can also query action items via the API to see pending, in-progress, completed, or overdue status. Q: What if the recording fails or has no transcript?
A: The system creates a meeting record with status “TRANSCRIPT_UNAVAILABLE” and notifies the organizer. No AI processing occurs. You can manually add notes using the manual meeting creation flow if needed.

​
Troubleshooting

Issue: I recorded a meeting but didn’t receive minutes or action items
Check that the meeting was recorded via tl;dv and that the webhook was successfully delivered. If the meeting appears with status “PROCESSING_FAILED,” the system attempted processing 3 times and failed. Contact support with the meeting ID for investigation. Verify that at least one meeting attendee is registered on the platform. Issue: Action items were assigned to the wrong person
The system matches owners by name or email with 90%+ accuracy for registered users. If someone was mismatched, you can manually reassign the action item. Consider updating user display names to match how they’re referenced in meetings (e.g., “Mike” vs “Michael”). Issue: The meeting was processed but not filed to my project vault
Vault filing requires at least one attendee to be linked to an active job. If no attendees match users with jobs, the system stores the meeting but doesn’t file it anywhere. Verify that meeting attendees have access to the relevant project. Issue: Very long meetings (3+ hours) seem truncated
The system processes up to 100,000 tokens per meeting (approximately 3 hours of content). Longer meetings are truncated to this limit, and a warning is logged. For critical long sessions, consider splitting into multiple recordings or manually adding key points that fell outside the processing window. Data Room - All meeting minutes are filed to your project’s Data Room under “Meeting Notes.” You can organize, annotate, and export meeting documentation alongside your other project artifacts. Navigate to /dataroomdetail to explore your vault. SupaSherpa - Your AI assistant can now reference meeting content when advising you. Ask about recent decisions, action item status, or specific discussions. SupaSherpa proactively reminds you of overdue commitments. Start a conversation at /supasherpa. AI Product Manager Signals - Meetings are scanned for risk indicators (timeline slippage, budget concerns, scope creep) and positive signals (milestone completion, customer validation). Your assigned sherpa receives alerts when the PM detects issues that need attention. Check your signals in the analytics dashboard.