CEO Board Deliberation Skill

Overview

The CEO Board Deliberation skill orchestrates multi-agent reasoning for strategic business decisions. It implements IndyDevDan's innovation of templated agent expertise with persistent learning.

Created: 2026-03-24 Framework: CleanExpo/NodeJS-Starter-V1 Agents: 9 specialized board members with persistent expertise

The Golden Example

To understand how the board works, review this complete example decision:

File: .pi/ceo-agents/briefs/_EXAMPLE-acquisition-offer.md

What it demonstrates:

How to structure a board brief
Expected deliberation process
Agent perspective patterns
How expertise files are updated
Output format for decision memos

Use this example to train new team members on the board's decision process.

Quick Start

1. Prepare Your Brief

Copy the template and fill it out:

cp .pi/ceo-agents/briefs/_TEMPLATE.md \
   .pi/ceo-agents/briefs/[YOUR-TOPIC]-[DATE].md

Required sections:

Situation — What's the context?
Stakes — What's at risk or possible?
Constraints — What are our limits?
Key Questions — What does the board decide?

2. Start the Board

/ceo-begin [YOUR-TOPIC]-[DATE]

The board will:

Review your brief
Deliberate with all 9 agents
Debate and challenge each other
Reach a decision
Update expertise files with their learning

3. Review the Decision

Find outputs at:

.pi/ceo-agents/
├── memos/[TOPIC]-[DATE].md        # Decision summary
├── conversations/[TOPIC]-[DATE].json # Full debate transcript
└── artifacts/[TOPIC]-[DATE].svg     # Visual arguments

The 9 Board Members

1. CEO

Perspective: Holistic business lens, execution reality Questions: Can we execute this? Does it align with strategy? How do we communicate it? Expertise File: .pi/ceo-agents/expertise/ceo.md

2. Revenue Agent

Perspective: Growth, unit economics, expansion Questions: How does this impact revenue? Is it capital-efficient? What's the market opportunity? Expertise File: .pi/ceo-agents/expertise/revenue.md

3. Product Strategist

Perspective: Feature prioritization, user value, roadmap Questions: Does this delight users? How does it fit the roadmap? What's the UX impact? Expertise File: .pi/ceo-agents/expertise/product-strategist.md

4. Technical Architect

Perspective: System design, feasibility, technical debt Questions: Is this technically sound? What's the implementation complexity? Scalability concerns? Expertise File: .pi/ceo-agents/expertise/technical-architect.md

5. Contrarian

Perspective: Risk identification, blindspots, second-order effects Questions: What could go wrong? What are we not seeing? When should we say "no"? Expertise File: .pi/ceo-agents/expertise/contrarian.md

6. Compounder

Perspective: Long-term leverage, compound effects, platform thinking Questions: What compound effects does this create? Will this give us an unfair advantage over time? Expertise File: .pi/ceo-agents/expertise/compounder.md

7. Custom Oracle

Perspective: Unite-Group domain expertise — 7 businesses, MACAS advisory, Brisbane/QLD AUS context Questions: What do domain experts know about this? Regulatory/compliance concerns (Privacy Act 1988, ATO)? Expertise File: .pi/ceo-agents/expertise/custom-oracle.md

8. Market Strategist

Perspective: Competitive positioning, market timing, go-to-market Questions: How does this position us competitively? Is timing right? Market adoption risk? Expertise File: .pi/ceo-agents/expertise/market-strategist.md

9. Moonshot

Perspective: Radical innovation, asymmetric bets, breakthrough possibilities Questions: Is there a bigger opportunity here? What's the moonshot play? Expertise File: .pi/ceo-agents/expertise/moonshot.md

How Agent Expertise Works

Each agent has a persistent expertise file that evolves:

Initial State

ceo.md
├── Role Description — What this agent brings
├── Core Perspectives — Their lens on problems
├── Decision History — Table of all decisions they've participated in
└── Learning Notes — What they've learned over time

After Each Deliberation

Agents update their files with:

Decision Record — Topic, position taken, outcome
Learning — What they learned from the debate
Pattern Recognition — How this fits with past decisions
Accuracy Tracking — How did their prediction fare?

Example Update

| Date | Topic | Position | Outcome | Notes |
|------|-------|----------|---------|-------|
| 24/03/2026 | Acquisition Offer | "Pursue, but negotiate harder" | TBD | Concerned about valuation, team retention |

Deliberation Quality Metrics

The board improves over time. Track:

Decision Quality — How well do decisions age?
Agent Agreement — Which agents tend to align?
Contrarian Accuracy — How often does the Contrarian spot real risks?
Compounder Insight — Which compound bets pay off?
Speed — How quickly does the board reach consensus?

Best Practices

For Brief Writers

Be Specific — Vague situations lead to weaker deliberations
Include Data — Concrete metrics help agents reason better
State Constraints — Be honest about limitations
Ask Real Questions — Don't lead the board toward a predetermined answer

For Board Users

Trust the Process — Diverse perspectives catch blindspots
Review Dissent — The Contrarian's concerns often matter most
Track Predictions — Come back and see how the board's predictions fared
Improve Expertise — Manually update agent files with industry insights

For Maintenance

Keep Expertise Current — Update agent files as patterns emerge
Archive Decisions — Keep decision history for reference
Learn from Outcomes — Track prediction accuracy
Customise the Oracle — Keep custom-oracle.md current with Unite-Group's evolving context

File Structure

.pi/
├── README.md                           # PI workspace overview
├── ceo-agents/
│   ├── briefs/
│   │   ├── _TEMPLATE.md                # Brief template (copy this)
│   │   ├── _EXAMPLE-acquisition-offer.md # Golden example
│   │   └── [TOPIC]-[DATE].md           # Your briefs go here
│   ├── deliberations/
│   │   └── [TOPIC]-[DATE].json         # Deliberation records
│   ├── memos/
│   │   └── [TOPIC]-[DATE].md           # Decision memos
│   ├── conversations/
│   │   └── [TOPIC]-[DATE].json         # Full transcripts
│   ├── artifacts/
│   │   └── [TOPIC]-[DATE].svg          # Visual arguments
│   └── expertise/
│       ├── ceo.md
│       ├── revenue.md
│       ├── product-strategist.md
│       ├── technical-architect.md
│       ├── contrarian.md
│       ├── compounder.md
│       ├── custom-oracle.md            # Unite-Group domain expertise
│       ├── market-strategist.md
│       └── moonshot.md
└── shared/
    └── context/
        └── _TEMPLATE.md                # Shared business context

Integration with Unite-Group

The CEO Board integrates with the Unite-Group workflow:

Briefs can reference Linear issues or business objectives
Memos inform ROADMAP.md updates
Expertise files accumulate knowledge about the 7 businesses over time
Custom Oracle provides AUS regulatory and MACAS advisory context

Related Commands

/ceo-begin — Start a board deliberation
/swarm-audit — Audit codebase before committing to a technical decision
/hey-claude — Load full project context before asking questions

Credits

Inspired by: IndyDevDan's agent scaling insights Insight: "If you template your engineering, your agents can do exactly what you did. This is the big advantage. When you're not creating prescriptions, workflows, and systems for your agents to repeat, you miss out on all the true leverage."

Last Updated: 24/03/2026 Status: Active

ceo-board