Analyze Transcript

What It Extracts

Requirements

• Explicitly stated requirements
• Implied requirements from discussions
• Feature requests and suggestions

Decisions

• Agreed-upon approaches
• Technology choices
• Scope decisions

Action Items

• Commitments made
• Follow-up tasks
• Assignments

Concerns

• Risks identified
• Issues raised
• Open questions

Extraction Patterns

Decision Markers

"We decided that..."
"The conclusion is..."
"Going with [option]..."
"Agreed: ..."
"Decision: ..."

Action Item Markers

"Action: [name] will..."
"TODO: ..."
"[Name] to follow up on..."
"We need to..."
"Next step: ..."

Requirement Discussions

"The system should..."
"Users need to be able to..."
"It must handle..."
"Requirements: ..."
"Feature request: ..."

Concern Markers

"I'm concerned about..."
"Risk: ..."
"Issue: ..."
"What if...?"
"Problem: ..."

Workflow

Step 1: Load Transcript

Read the transcript file(s) using the Read tool.

Step 2: Identify Speakers

If --speakers provided, use the mapping. Otherwise, attempt to identify speakers from patterns like:

• [Name]: or Name:
• **Name**:
• Speaker labels in transcript format

Step 3: Extract Content

Spawn document-miner agent with transcript-specific extraction patterns.

Extract:

1. Requirements (explicit and implicit)
2. Decisions
3. Action items
4. Concerns and risks

Step 4: Attribute and Weight

For each extraction:

• Note the speaker (if identifiable)
• Apply weighting based on speaker role
• Record surrounding context

Step 5: Save and Report

Save to .requirements/{domain}/transcripts/ Report summary of findings.

Speaker Weighting

speaker_weights:
  product_owner: high    # Business requirements
  technical_lead: high   # Technical constraints
  stakeholder: high      # Domain requirements
  developer: medium      # Implementation insights
  designer: medium       # UX requirements
  unknown: low           # Needs validation

Examples

Basic Transcript Analysis

/requirements-elicitation:analyze-transcript ./meetings/sprint-planning.md

Output:

Analyzing: sprint-planning.md

Speakers identified:
- PM (ProductManager): 45 statements
- TL (TechLead): 32 statements
- DEV (Developer): 28 statements

Extraction Results:

REQUIREMENTS (8):
- REQ-TR-001: "Users shall receive email notifications for order status" [PM, High]
- REQ-TR-002: "API response time must be under 200ms" [TL, High]
...

DECISIONS (3):
- DEC-001: "Use PostgreSQL for primary database" [TL]
- DEC-002: "Target launch date: Q2 2025" [PM]
...

ACTION ITEMS (5):
- ACT-001: "TL to create database schema document"
- ACT-002: "PM to get stakeholder sign-off"
...

CONCERNS (2):
- CONCERN-001: "Third-party API reliability" [TL]
- CONCERN-002: "Training timeline for new features" [PM]

Saved to: .requirements/sprint-planning/transcripts/TR-sprint-planning.yaml

With Speaker Mapping

/requirements-elicitation:analyze-transcript ./calls/client-call.txt \
  --speakers "JD:ClientCEO,SM:SalesManager,PM:ProductManager" \
  --domain "enterprise-deal"

Output:

Analyzing: client-call.txt

Speaker Mapping Applied:
- JD → ClientCEO (weight: high)
- SM → SalesManager (weight: medium)
- PM → ProductManager (weight: high)

REQUIREMENTS from ClientCEO (5):
- REQ-TR-001: "Must integrate with SAP" [JD, High confidence]
- REQ-TR-002: "Support 50,000 concurrent users" [JD, High confidence]
...

REQUIREMENTS from ProductManager (3):
- REQ-TR-006: "Dashboard customization needed" [PM, High confidence]
...

Saved to: .requirements/enterprise-deal/transcripts/TR-client-call.yaml

Multiple Transcripts

/requirements-elicitation:analyze-transcript ./meetings/*.md --domain "q1-planning"

Output:

Found 4 transcripts matching pattern

Processing:
1. kickoff-meeting.md ........ 12 requirements, 4 decisions
2. technical-review.md ....... 8 requirements, 6 decisions
3. stakeholder-feedback.md ... 15 requirements, 2 decisions
4. sprint-planning.md ........ 5 requirements, 3 decisions

Cross-Transcript Analysis:
- Total requirements: 40
- Duplicates detected: 6 (consolidated to 34)
- Conflicting statements: 2 (flagged for review)

Summary saved to: .requirements/q1-planning/transcripts/summary.yaml

Output Format

Saved YAML Structure

transcript_analysis:
  file: "sprint-planning.md"
  analyzed_date: "2025-12-25T14:30:00Z"
  domain: "{domain}"