Capture Meeting

Capture Meeting

Process meeting notes or transcript to extract decisions, commitments, and insights.

Trigger

• "Here's a transcript from [client/person]"
• "Process these meeting notes"
• "Here are my notes from the call with [person]"
• "Capture this meeting"
• /capture-meeting

Input

User provides one of:

• Full transcript (from Otter, Granola, etc.)
• Meeting notes (manual)
• Voice memo summary
• Memory/verbal recap

Processing Steps

1. File the Source Material (MANDATORY)

Always file the raw transcript/notes FIRST. This is not optional. Source preservation creates provenance: every extracted fact can trace back to where it came from.

Call the memory_file MCP tool with:

• filename: "YYYY-MM-DD-[person]-[topic].md"
• source_type: "transcript"
• summary: "Brief 1-line summary of the meeting"
• about: ["participant1", "participant2"]
• content: The FULL raw transcript/notes text (do not summarize)

The file is automatically routed to the right folder:

• people/sarah-chen/transcripts/2026-02-04-kickoff.md
• clients/acme-corp/transcripts/2026-02-04-quarterly.md

Even for brief notes: If the user shared more than a few sentences, file it. Better to have it than wish you did.

2. Identify Participants

• Who was in the meeting?
• Which person files to update?
• Any new people to track?

3. Extract Key Information (Agent-Accelerated)

Preferred: Dispatch Document Processor for extraction. Instead of composing memory operations manually (which takes 2+ minutes of thinking time), dispatch the Document Processor agent (Haiku) with the transcript content and extraction_type: "memory_operations". The agent returns ready-to-store operations in ~10-20 seconds.

Agent pipeline workflow:

1. Dispatch Document Processor (Haiku) with:
   - The full transcript text
   - extraction_type: "memory_operations"
   - Context: participant names, meeting topic, date

2. Agent returns memory_operations[] array with:
   - Facts, preferences, observations
   - Commitments with deadlines
   - Entity definitions
   - Relationship links

3. Review agent output (judgment layer):
   - Verify commitment wording is accurate
   - Check importance scores are reasonable
   - Confirm entity names match existing entities
   - Adjust or remove any questionable extractions

4. Call the `memory_batch` MCP tool with the reviewed operations array

Fallback: Manual extraction (use when agent is unavailable or for very short notes)

Decisions Made:

• What was decided?
• Who made the decision?
• Any conditions or context?

Commitments Created:

• What did you promise? (→ context/commitments.md)
• What did they promise? (→ context/waiting.md)
• Deadlines (explicit or implied)

Blockers Surfaced:

• What's in the way?
• Who can unblock?

Sentiment Signals:

• Enthusiasm, concern, resistance
• Energy level
• Relationship health indicators

Key Topics:

• Main themes discussed
• Important context shared

4. Link Provenance

After extracting memories (facts, commitments) via the memory_batch or memory_remember MCP tools:

Call the memory_file MCP tool with the memory_ids parameter set to the IDs of the memories you extracted. This links the stored transcript to the memories extracted from it, creating the provenance chain: memory -> document -> file on disk.

Now the user can ask "where did you learn that Sarah prefers async communication?" and you can point to the exact transcript.

5. Downstream Updates

After extracting information, propagate changes to the files that depend on it. These updates ensure that summaries stay in sync with source material.

5a. Update person files

For each participant in the meeting:

1. Check if people/[name-slug].md exists
2. If it exists:
   • Update "Last Contact" date to today
   • Add this meeting to "Our History" or interaction log
   • Update "Current Context" if new information was shared
   • Add any new commitments to the person's section
3. If it does not exist and this person seems important (mentioned multiple times, has commitments, or has a working relationship):
   • Offer to create a new person file: "I'd like to create a file for [person]. They [reason]. Should I?"

5b. Update commitment and waiting files

• Add new commitments to context/commitments.md (ask for confirmation on wording and deadline)
• Add new waiting items to context/waiting.md
• If memory MCP tools are available, also store via the memory_remember MCP tool

5c. Update workspace files (if applicable)

Check if this meeting belongs to an active workspace:

1. Does the meeting topic match a workspace in workspaces/?
2. Is a meeting participant associated with a workspace project?

If yes:

• File the meeting notes in workspaces/[slug]/meetings/
• Update workspaces/[slug]/Dashboard.md if the meeting changed project status or phase
• If the meeting created new deliverables or items, add them to the relevant workspace subdirectory

If no workspace match, file in the standard location per the document store routing.

6. Synthesize

Create a summary that captures:

• What happened (brief)
• What was decided
• What's next (actions)
• How it went (sentiment)

Output Format

**📋 Meeting Capture: [Meeting Name/Person]**
### [Date]

**Attendees:** [Names]
**Duration:** [Approximate]
**Context:** [Brief — what was this meeting about?]

### 📝 Summary
[2-3 sentence overview of what happened]

### 🔨 Decisions Made
- [Decision] — decided by [who]
- [Decision]

### ✅ Action Items

**You:**
- [ ] [Action] — by [date]
- [ ] [Action] — by [date]

**Them:**
- [ ] [Action] — by [date]

### 💬 Key Discussion Points
- [Point 1]
- [Point 2]
- [Point 3]

### 🌡️ Sentiment
[Brief read on how the meeting went, relationship health]

### 📂 Updates Made

**Person files:**
- Updated [person]'s last contact and history
- Added [commitment] to [person]'s commitments section
- ? Create file for [new person]? They [reason]. (Your call.)

**Commitments added:**
- "[Commitment]" — due [date] (added to tracking)
- Waiting on "[item]" from [person] (added to waiting)

**Workspace:** [if applicable]
- Meeting filed in workspaces/[slug]/meetings/
- Dashboard updated: [what changed]

*Anything I should adjust?*

*Meeting notes saved to: [location]*

---

Judgment Points

Proceed automatically with:

• Updating last contact dates in person files (factual, low-risk)
• Adding meeting to history tables in person files (factual, low-risk)
• Filing meeting notes in workspace directories (organizational, low-risk)

Ask for confirmation on:

• Adding commitments (user must own promises)
• Adding waiting items (setting expectations)
• Creating NEW person files (new entity in the system)
• Updating sentiment in person files (subjective)
• Changing project phase/status in workspace Dashboard (consequential)
• Flagging concerns (interpretation required)
• File location (if ambiguous)

Quality Checklist

• Raw transcript/notes filed (memory_file MCP tool called with full content)
• Memories linked to source document (provenance chain complete)
• Every action item has an owner
• Every commitment has a deadline (even approximate)
• Sentiment signals noted but not over-interpreted
• Summary is actionable, not just descriptive
• Related person files flagged for update
• No unexplained jargon or unclear references
• All markdown tables render correctly (header, separator, and data rows on separate lines)
• Person files updated for all participants with existing files
• Workspace files updated if meeting belongs to an active project
• No stale counts left in any summary file that was updated

Tone

• Efficient — respect user's time
• Accurate — don't add or assume
• Helpful — surface the useful parts
• Action-oriented — what needs to happen next