The Quorum — Build Your Own Board of Directors

The Prompt Sequence

The Starting Point

Introduce the concept and ask Claude to generate 20 candidate board members from your personal OS. You select up to 8 across three archetypes.

MentorSponsorChallenger

Use this prompt

Hi Claude, I'd like to build out a personal board of directors. It's a small group of trusted advisors — real people, dead or alive — that you personally select to help guide your career and big decisions. A strong board has three types: a Mentor, a Sponsor, and a Challenger. Based on my personal OS, which famous people should be on my board? Max 8. Give me 20 options and summarize each in 7 words or less.

Build the Command

Turn your 8 board members into a live /journey-bod command. All 8 run as parallel agents simultaneously using a Ralph Loop, debate in up to 3 rounds, then converge on a recommendation.

Ralph LoopParallel AgentsConsensus

Use this prompt

Now build me a /journey-bod command I can run anytime I have a question or decision I'm wrestling with. Pull in all 8 perspectives at once using a Ralph Loop — run all 8 board members as parallel agents simultaneously. Then let them react in up to 3 rounds of debate. One synthesized recommendation. If they can't agree after 3 rounds, [Chairman name] synthesizes a clear recommendation from the full deliberation.

Add Context and Memory

Give the board awareness of its own history. Each session saves as a unique file. Future sessions open from the last meeting by default, or you can point to any session you want.

Session HistoryContinuityConfirmation

Use this prompt

A few things to add: before the board starts, tell me what it thinks I'm asking and let me confirm. Instead of overwriting the same file every time, save each session as its own file. By default, when I reconvene, pick up from the most recent session — but let me point to a specific one if I want.

Make the Board Push Back

Remove the sycophancy. If the board believes it's right and you challenge them, they hold the line. Add a round-count to every session so you know how contested the decision really was.

ConvictionIntegrityRound Count

Use this prompt

I want to make sure the board isn't just telling me what I want to hear. If I push back, I don't want them to roll over. If they believe they're right, hold the line — be smart, be respectful, but don't cave. Also show me how many rounds it took to reach a decision.

Make It Cheaper to Run

Cut token cost without losing what makes the board useful. Tighter word limits per round, one agent per member across all rounds, summaries instead of full transcripts, and automatic second-brain logging.

Token EfficiencyAuto-SaveSummaries

Use this prompt

Help me make it more token efficient without losing what makes it useful: 75 words max in Round 1, 50 in Round 2, 30 in Round 3. One agent per board member across all 3 rounds. Between rounds, each agent gets only a one-line summary of others' positions. Don't show me back-and-forth — process in memory and surface only the final synthesis. Auto-save to my second brain.

Live Session Example

Here is a real board session, start to finish. The question, the three-round deliberation, and the Chairman's synthesized recommendation — all from a single /journey-bod invocation.

Phase 01 The Question

jordan@bod — ~/TrainingCourse

❯ /journey-bod

The Question

I'm energized by building AI agents, but I spent two weeks of tokens building a morning brief that surfaces consolidated tasks, open commitments, and nuanced waiting items across disparate sources.

It has real utility — but was the investment worth it?

Help me think about when to time-box and token-box my efforts based on utility, habit-building likelihood, and learnings.

8 board members standing by

↓

Phase 02 The Deliberation — Ralph Loop, 3 Rounds

Ralph Loop — 3 Rounds in Parallel

Round 01 — Independent

Buffett No pre-investment checklist. Tinkering ≠ strategy. PAUSE

Drucker Two weeks displaced what? Real cost = opportunity cost. PAUSE

Ogilvy Write the behavior change, not the feature, before building. PAUSE

Jobs Built comprehensive before testing smallest viable. PIVOT

Graham Build the dumbest version in one day. Validate the habit first. PIVOT

Hopper It works. Extract the lesson and keep moving. PROCEED

Franklin 3-axis test: autonomous? compounding? survives a week idle? PAUSE

Oprah Tuition or avoidance? Who are you becoming by building this? PAUSE

5 PAUSE

2 PIVOT

1 PROCEED

Round 02 — Cross-Pollination

Buffett Franklin's 3-axis test + 3-day hard cap before prototype. PAUSE

Drucker Behavior first, minimum build, then 14-day measurement. PAUSE

Ogilvy Oprah's tuition/avoidance reframe changes the calculus. PAUSE

Jobs PAUSE wants more gates. I want smaller builds. Not the same. PIVOT

Graham Day-one signal beats a 14-day discipline measurement. PIVOT

Hopper Framework emerges from doing, not deliberating. PROCEED

Franklin Enthusiasm ≠ system. Build the feedback loop first. PAUSE

Oprah Day-one usage is a proxy, not a truth. Some tools reveal slowly. PAUSE

5 PAUSE

2 PIVOT

1 PROCEED

Round 03 — Convergence

Buffett 3 numbers before any build: value, frequency, stop condition. PAUSE

Drucker Behavior change + 14-day measure. That is the whole system. PAUSE

Ogilvy Research the behavior. Build the minimum. Honor the utility. PAUSE

Jobs Accept PAUSE only if max 3-day build rule is the center. D&C

Graham Accept PAUSE only if day-one usage is the primary signal. D&C

Hopper Accept PAUSE only if a written lesson log keeps builds moving. D&C

Franklin 3-axis test + build log, reviewed quarterly. PAUSE

Oprah Tuition or avoidance before. Behavior check after 30 days. PAUSE

5 PAUSE

3 D&C

No 6/8 → Chairman

3 rounds complete

Deadlock — Chairman activated

↓

Phase 03 The Recommendation

Chairman's Synthesis — Warren Buffett

Board Recommendation — PAUSE

Step	Action
Before	Write the behavior change, the day-one test, and the stop condition — 20 minutes max
During	3-day cap before first use; build the smallest version that tests the behavior
After 30d	Franklin's three-axis check + Oprah's tuition-or-avoidance question

Jobs & Graham's condition preserved: the day-one signal — did you open it the first morning without reminding yourself? — is the single most predictive data point. Everything else is measuring discipline, not utility.

PAUSE

3 rounds

Chairman's Synthesis

2026-04-13