/conclave Command

Multi-agent deliberation system for strategic decisions with evidence-based debate, critical review, and synthesis

/conclave - Council Deliberation

Executes a complete council session: multi-agent debate followed by meta-evaluation by Critic, Devil’s Advocate, and Synthesizer agents.

Syntax

/conclave [QUESTION]

QUESTION · string · required

Strategic decision or question requiring multi-perspective analysis

Example: “Should we increase closer commission from 10% to 15%?”

What It Does

Conclave orchestrates a 4-phase deliberation:

┌───────────────────────────────────────────────────────────────────────────
│                          CONCLAVE FLOW                                     │
├───────────────────────────────────────────────────────────────────────────┤
│  Phase 0: Constitutional Principles     [Foundation]                       │
│  Phase 1: Multi-Agent Debate            [Cargo agents deliberate]         │
│  Phase 2: Methodological Critic         [Process evaluation]              │
│  Phase 3: Devil's Advocate              [Challenge assumptions]           │
│  Phase 4: Synthesis                     [Final recommendation]            │
└───────────────────────────────────────────────────────────────────────────┘

Anti-Loop Rule: Conclave runs ONCE per query. If confidence < 60%, escalates to human decision (does NOT re-run).

Phase Breakdown

Phase 0: Constitutional Foundation

Purpose: Establish governing principles for deliberation

Principle 1: Empiricism

Mandate:
  "Decisions based on DATA, not opinions or intuitions"

Requirements:
  - All claims must cite sources
  - Use concrete numbers
  - Reference chunk_ids for evidence

Example:
  Good: "Close rate drops 40% ^[CG003_089]"
  Bad: "Close rate probably drops significantly"

Principle 2: Pareto (80/20)

Mandate:
  "Seek 20% of actions that generate 80% of results"

Focus:
  - What has highest leverage?
  - Which option requires least effort for most gain?
  - Avoid "busy work" solutions

Example:
  Question: "Should we hire 5 or 10 closers?"
  Pareto lens: "Which number hits 80% of revenue goal?"

Principle 3: Inversion

Mandate:
  "Before asking 'what to do', ask 'what would cause failure'"

Process:
  1. Identify ways decision could fail
  2. Design around failure modes
  3. Build in safeguards

Example:
  Decision: "Launch new product"
  Inversion: "What would make this product fail?"
  - No market demand
  - Wrong pricing
  - Poor onboarding
  → Address each before launch

Principle 4: Antifragility

Mandate:
  "Prefer options that benefit from volatility and stress"

Evaluation:
  - Which option gets stronger under pressure?
  - What improves with chaos?
  - Where is optionality preserved?

Example:
  Option A: Fixed-price contract
  Option B: Performance-based pricing
  
  Under stress:
  A: No benefit from volatility
  B: Aligns incentives, benefits from success
  
  → Choose B (antifragile)

Hierarchy:

CONSTITUTION
  ↓
PROTOCOLS
  ↓
AGENT INSTRUCTIONS

If any agent violates constitutional principles, the Critic will flag it in Phase 2.

Phase 1: Multi-Agent Debate

Purpose: Gather perspectives from relevant cargo agents

Agent Selection:
  Based on question domain:
  
  Commission question → [CRO, CFO, SALES-MANAGER, CLOSER]
  Hiring decision → [CRO, SALES-MANAGER, COO]
  Pricing change → [CRO, CFO, CLOSER]
  Process improvement → [COO, relevant role]

Debate Format:
  1. Each agent presents position
  2. Cites evidence (chunk_ids)
  3. Identifies risks and trade-offs
  4. States confidence level

Example Debate Output:

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
PHASE 1: CARGO AGENT DEBATE
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

QUESTION: "Should we increase closer commission from 10% to 15%?"

┌─ CRO (Chief Revenue Officer) ───────────────────────────────────────────────────────
│                                                                            │
│  POSITION: YES, with conditions                                           │
│                                                                            │
│  REASONING:                                                                │
│  Higher commission attracts A-players. Industry standard is 12-15%        │
│  ^[AH002_045]. Our current 10% makes us uncompetitive for top talent.      │
│                                                                            │
│  EVIDENCE:                                                                 │
│  • Cole Gordon data: 15% commission → 30% better retention [CG001_089]  │
│  • Cost per acquisition drops when closers stay longer [AH003_023]       │
│                                                                            │
│  CONDITIONS:                                                               │
│  • Tie to performance: 12% base + 3% accelerator at 80% quota           │
│  • Phase in over 2 quarters to test impact                               │
│                                                                            │
│  CONFIDENCE: 75%                                                           │
│                                                                            │
└────────────────────────────────────────────────────────────────────────────┘

┌─ CFO (Chief Financial Officer) ────────────────────────────────────────────────────
│                                                                            │
│  POSITION: CONDITIONAL YES                                                │
│                                                                            │
│  REASONING:                                                                │
│  5% increase = ~$250k/year additional cost at current volume. Need to     │
│  ensure ROI justifies expense.                                            │
│                                                                            │
│  FINANCIAL MODEL:                                                          │
│  Current:  10% commission on $5M ARR = $500k                              │
│  Proposed: 15% commission on $5M ARR = $750k (+$250k)                     │
│                                                                            │
│  BREAK-EVEN:                                                               │
│  Need 5% revenue increase to justify cost ($250k)                         │
│  = $5.25M ARR (easily achievable if retention improves)                   │
│                                                                            │
│  CONDITIONS:                                                               │
│  • Monthly KPI tracking: retention, close rate, pipeline velocity        │
│  • Revert if ROI negative after 6 months                                  │
│                                                                            │
│  CONFIDENCE: 70%                                                           │
│                                                                            │
└────────────────────────────────────────────────────────────────────────────┘

┌─ SALES-MANAGER ────────────────────────────────────────────────────────────────
│                                                                            │
│  POSITION: STRONG YES                                                     │
│                                                                            │
│  REASONING:                                                                │
│  Recruiting is brutal right now. Lost 3 candidates last month to          │
│  competitors offering 15%. Current team morale suffering.                 │
│                                                                            │
│  OPERATIONAL IMPACT:                                                       │
│  • Recruiting cycle: 45 days → ~25 days (based on market intel)        │
│  • Churn: Current 40% annually, expect drop to ~25% [CG002_034]          │
│  • Training ROI improves when closers stay longer                         │
│                                                                            │
│  TEAM FEEDBACK:                                                            │
│  Asked top 3 closers: all said they'd consider external offers           │
│  without increase. Risk of losing A-players.                              │
│                                                                            │
│  CONFIDENCE: 85%                                                           │
│                                                                            │
└────────────────────────────────────────────────────────────────────────────┘

┌─ CLOSER ──────────────────────────────────────────────────────────────────────────
│                                                                            │
│  POSITION: YES (obviously)                                                │
│                                                                            │
│  REASONING:                                                                │
│  Every closer I know in our niche is making 15%+. We're leaving money     │
│  on the table and it affects motivation.                                  │
│                                                                            │
│  REALITY CHECK:                                                            │
│  • Our close rate is 65% (above industry 45% ^[CG001_023])              │
│  • We EARN the higher commission through performance                      │
│  • Better closers = less sales cycle friction = happier customers        │
│                                                                            │
│  PREFERENCE:                                                               │
│  Accelerator model > flat 15%. Rewards performance, not just showing up.  │
│                                                                            │
│  CONFIDENCE: 90%                                                           │
│                                                                            │
└────────────────────────────────────────────────────────────────────────────┘

Phase 2: Methodological Critic

Purpose: Evaluate the quality of the debate process (not the merit of positions)

Critic Evaluates:
  ✓ Were premises clearly stated?
  ✓ Is evidence traceable (chunk_ids)?
  ✓ Is logic consistent?
  ✓ Were alternative scenarios considered?
  ✓ Were conflicts/tensions resolved?

Score: 0-100
  - Premises declared: 0-20 pts
  - Evidence traceable: 0-20 pts
  - Logic consistent: 0-20 pts
  - Alternatives considered: 0-20 pts
  - Conflicts resolved: 0-20 pts

Example Output:

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
PHASE 2: METHODOLOGICAL CRITIC
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

QUALITY SCORE: 82/100

Breakdown:
  • Premises declared:      18/20  (Clear problem statement, context)
  • Evidence traceable:      16/20  (Most claims cited, 2 missing chunk_ids)
  • Logic consistent:        20/20  (No contradictions)
  • Alternatives considered:  15/20  (Missed "status quo" scenario)
  • Conflicts resolved:       13/20  (CRO/CFO tension not fully addressed)

GAPS IDENTIFIED:
  • Missing: What if we do nothing? (status quo scenario)
  • Missing: Sales-Manager claim "lost 3 candidates" - no chunk_id
  • Tension: CRO wants immediate, CFO wants phased - needs resolution

RECOMMENDATION: APPROVE with minor revisions
  Address identified gaps before final decision.

Phase 3: Devil’s Advocate

Purpose: Challenge assumptions and identify hidden risks

Advocate Identifies:
  • Most fragile premise
  • Main risk NOT discussed
  • 12-month regret scenario (what could go wrong)
  • Ignored alternative

Example Output:

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
PHASE 3: DEVIL'S ADVOCATE
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

MOST FRAGILE PREMISE:
  "Higher commission attracts A-players"
  
  Challenge:
  Assumes commission is PRIMARY motivator. What if A-players care more
  about:
    • Product quality (do they believe in what they're selling?)
    • Career growth (path to leadership?)
    • Work environment (team culture, autonomy?)
  
  Risk:
  We spend $250k/year and still lose top talent because compensation
  wasn't the real problem.

MAIN RISK NOT DISCUSSED:
  ADVERSE SELECTION
  
  What if:
  15% commission attracts MERCENARIES who optimize for short-term
  commissions over customer success.
  
  Impact:
    • Close rate goes up (good)
    • Churn goes up (bad)
    • LTV drops (very bad)
  
  This is a 12-24 month delayed consequence. By the time we notice,
  damage is done.

12-MONTH REGRET SCENARIO:
  It's March 2027. We increased commission to 15%.
  
  What happened:
  • Hired 5 new closers attracted by high commission
  • 3 of them hit quota... by overselling and setting wrong expectations
  • Customer churn spiked from 15% to 28%
  • NPS dropped from 45 to 28
  • Now spending more on acquisition AND dealing with reputation damage
  
  CFO's nightmare: "We optimized for top-line, destroyed bottom-line."

ALTERNATIVE IGNORED:
  NON-MONETARY COMP
  
  What if we:
  • Keep 10% commission
  • Add equity/profit-sharing for top performers
  • Create clear path to VP Sales (career track)
  • Invest $100k in training/coaching instead of $250k in commission
  
  Benefits:
  • Attracts long-term thinkers, not mercenaries
  • Aligns with company success (equity)
  • Costs less
  
  This wasn't even mentioned in debate.

Phase 4: Synthesis

Purpose: Integrate feedback and produce final recommendation

Synthesizer Produces:
  • Clear decision
  • Confidence score (0-100%)
  • Modifications from feedback
  • Residual risks + mitigations
  • Next steps (who, what, when)
  • Reversal criteria

Example Output:

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
PHASE 4: FINAL SYNTHESIS
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

DECISION: YES - Increase commission with safeguards

CONFIDENCE: 78%

Justification:
  Consensus among 3/4 agents (CRO, CFO, Sales-Manager, Closer).
  Financial model shows break-even is achievable.
  Recruiting/retention pain is real and urgent.
  Devil's Advocate concerns are valid but addressable.

MODIFICATIONS (based on Critic + Advocate feedback):

  FROM (initial proposal):
    Flat 15% commission immediately
  
  TO (revised):
    TIER 1: 12% base commission
    TIER 2: +3% accelerator at 80% quota
    TIER 3: +2% additional at 100% quota with NPS > 40
    
    → Top performers earn 17% (better than market)
    → Addresses adverse selection (NPS gate)
    → Costs less upfront ($150k vs $250k)

  PHASE-IN:
    Q1: Announce change, set expectations
    Q2: Implement Tier 1 (12%)
    Q3: Add Tier 2 (accelerator)
    Q4: Full structure (with NPS gate)

RESIDUAL RISKS + MITIGATIONS:

  RISK 1: Adverse selection (mercenaries)
    Mitigation:
      • NPS gate on top tier (customer success must stay high)
      • 6-month probation period
      • Monthly churn monitoring
    Owner: Sales-Manager + CFO
  
  RISK 2: Cost overrun if volume spikes
    Mitigation:
      • Cap total commission pool at 18% of revenue
      • Quarterly review with CFO
    Owner: CFO
  
  RISK 3: Existing team feels "baited" if accelerator is hard to hit
    Mitigation:
      • 80% quota threshold is achievable (current avg: 85%)
      • Transparent communication about tiers
    Owner: Sales-Manager

NEXT STEPS:

  1. Draft compensation policy document
     Owner: CRO + CFO
     Deadline: End of week
  
  2. Model 3 scenarios (conservative, base, aggressive revenue)
     Owner: CFO
     Deadline: 5 days
  
  3. Socialize with top 3 closers (get feedback)
     Owner: Sales-Manager
     Deadline: 1 week
  
  4. Board approval (if needed for budget)
     Owner: CEO
     Deadline: 2 weeks
  
  5. Company-wide announcement
     Owner: CRO
     Deadline: 3 weeks

REVERSAL CRITERIA:

  Pause or revert if ANY of:
    • Churn rate increases > 5% absolute within 6 months
    • NPS drops below 35
    • Cost as % of revenue exceeds 18%
    • Customer complaints about "overselling" spike (>10/month)
  
  Review: Monthly for 6 months, then quarterly

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

CONSENSUS ACHIEVED
Decision ready for executive approval.

Low Confidence Outcome

If confidence < 60%, escalates to human:

░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░
[COUNCIL: DECISION INCONCLUSIVE]
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

⚠️  CONFIDENCE: 52% - BELOW THRESHOLD OF 60%

TYPE OF UNCERTAINTY:
  [✓] Insufficient data
  [ ] Irresolvable conflict between roles
  [ ] Outside scope of available knowledge

OPTIONS FOR HUMAN DECISION:

OPTION A: Increase to 15% immediately
  Trade-off: Speed vs risk
  Defendedby: CRO, Sales-Manager, Closer
  Evidence: [CG001_089, AH002_045]

OPTION B: Phased approach (12% → 15% over 2 quarters)
  Trade-off: Safety vs market competitiveness
  Defended by: CFO
  Evidence: [Financial model]

OPTION C: Non-monetary comp (equity, career path)
  Trade-off: Long-term alignment vs immediate fix
  Suggested by: Devil's Advocate
  Evidence: [No direct evidence, hypothesis]

OPTION D: Gather more data
  What's needed:
    • Survey top 10 closers: what would make them stay?
    • Benchmark 5 competitors: actual comp packages
    • Model churn impact with real numbers
  How to obtain: 2-week research sprint

⚠️  This case requires HUMAN DECISION.
The Council is NOT recommending any option.
░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░░

Examples

Commission Change

/conclave "Should we increase closer commission from 10% to 15%?"

When to Use

Good Use Cases

✓ Strategic decisions - Affects money, structure, risk ✓ Multiple stakeholders - Different roles have valid perspectives ✓ Evidence exists - Knowledge base has relevant data ✓ Reversible - Can undo if wrong

Poor Use Cases

✗ Tactical execution - “Which CRM should we use?” ✗ No evidence - Knowledge base lacks relevant data ✗ Urgent - Need decision in < 1 hour ✗ Highly subjective - “Should logo be blue or green?”

Agent Roles

Methodological Critic

Evaluates process quality, not merit of ideas. Ensures rigor.

Devil's Advocate

Challenges assumptions, identifies blind spots, surfaces risks.

Synthesizer

Integrates feedback, resolves tensions, produces final decision.

Cargo Agents

CRO, CFO, Sales-Manager, etc. Bring domain expertise.

Key Principles

1. Evidence-Based

All claims must cite sources:

Good:
  "Close rate drops 40% without qualification ^[CG003_089]"

Bad:
  "Close rate probably drops without qualification"

2. Constitutional Governance

Principles override agent opinions:

CONSTITUTION > PROTOCOLS > AGENT INSTRUCTIONS

3. One-Pass Rule

Conclave runs once. No loops. If inconclusive, escalate to human.

4. Process vs Merit

Critic evaluates HOW agents deliberated, not WHAT they concluded.

5. Adversarial Testing

Devil’s Advocate actively seeks to break the conclusion.

Troubleshooting

”No relevant agents found”

Issue: Question doesn’t map to any cargo agents

Solution:

# Check available agents
ls agents/cargo/

# Ensure question matches agent domains
# Example: Marketing questions need CMO agent

“Insufficient evidence”

Issue: Knowledge base lacks data for question

Solution:

# Ingest relevant materials first
/ingest expert-video.mp4
/process-jarvis "inbox/EXPERT/video.txt"

# Then retry conclave
/conclave "your question"

“Confidence always < 60%”

Issue: Agents consistently uncertain

Solution: This indicates:

• Genuinely ambiguous decision
• Missing critical data
• Question outside agent expertise

Human decision required.

Next Steps

Agent System

Understand cargo agents and mind clones

Evidence Tracing

How chunk_ids enable traceability

Constitutional Principles

Deep dive into governing principles

Decision Framework

When to use Conclave vs other tools