Have you ever asked Claude (or ChatGPT) a question, it responds with some ludicrous answer, you push back, and it concedes with “You’re right” (😪)
Claude is the Yes Man none of us need.
Contract trade-offs, investment decisions, whether to build the workflow or buy it — don't have one clean answer. Oftentimes, there are multiple stakeholder in a room evaluating the decision and providing their insight based off their background, role, and context. This is where the useful information lives.
A "Claude Council" is a prompting structure that recreates the room: you assign the model several deliberately different advisor lenses, instruct them to respond independently, then appoint a chair to reconcile the disagreement into a decision. It’s not a new Anthropic product, but rather just a way of asking.
The mechanic
Like a PB&J, a council prompt is made of three key ingredients:
The decision, phrased neutrally. Not "why is this headcount investment the right move"
The roles. Either you pick them, or you tell Claude to pick 4-6 and state each one's lens before it starts talking (see example prompt below)
The output. Remember: Independent opinions first. Chair synthesis after. Never (ever) skip straight to the summary — that's where the disagreement gets averaged away.
Why bother
Most questions don't need this. A council is helpful when you’re deciding on ambiguous, high-judgment calls.
Take a price increase for a customer at renewal. One advisor stress-tests churn risk. Another checks whether the unit economics and value metric actually support it. A third goes looking for a bundling play not considered. The chair turns that into a recommendation, stated risks, and next steps versus an answer which omits real tradeoffs without surfacing first.
One caveat: a council doesn't make anything true. It's a reasoning structure, not a be-all-end-all. You earn the big bucks by still checking the contract language, the actual numbers, and the policy fine print yourself. The council just tells you where to look harder and what you may have missed.
The minimum viable prompt
You are my decision council.
Evaluate the question below independently from five roles:
1. Contrarian — identify failure modes and disconfirming evidence.
2. First-principles thinker — identify the true problem, assumptions, and constraints.
3. Expansionist — identify upside and overlooked strategic options.
4. Outsider — provide a fresh perspective and challenge internal assumptions.
5. Executor — recommend concrete actions, owners, and sequencing.
Rules:
- State assumptions explicitly.
- Distinguish facts, assumptions, and recommendations.
- Encourage disagreement; do not manufacture consensus.
- Flag what requires external verification.
Then act as Chair:
- Recommend one path.
- State the principal risk.
- Specify the first three actions and the evidence needed to validate them.
Question: [insert decision]
You are my decision council.
Evaluate the question below independently from five roles:
1. Contrarian — identify failure modes and disconfirming evidence.
2. First-principles thinker — identify the true problem, assumptions, and constraints.
3. Expansionist — identify upside and overlooked strategic options.
4. Outsider — provide a fresh perspective and challenge internal assumptions.
5. Executor — recommend concrete actions, owners, and sequencing.
Rules:
- State assumptions explicitly.
- Distinguish facts, assumptions, and recommendations.
- Encourage disagreement; do not manufacture consensus.
- Flag what requires external verification.
Then act as Chair:
- Recommend one path.
- State the principal risk.
- Specify the first three actions and the evidence needed to validate them.
Question: [insert decision]
You are my decision council.
Evaluate the question below independently from five roles:
1. Contrarian — identify failure modes and disconfirming evidence.
2. First-principles thinker — identify the true problem, assumptions, and constraints.
3. Expansionist — identify upside and overlooked strategic options.
4. Outsider — provide a fresh perspective and challenge internal assumptions.
5. Executor — recommend concrete actions, owners, and sequencing.
Rules:
- State assumptions explicitly.
- Distinguish facts, assumptions, and recommendations.
- Encourage disagreement; do not manufacture consensus.
- Flag what requires external verification.
Then act as Chair:
- Recommend one path.
- State the principal risk.
- Specify the first three actions and the evidence needed to validate them.
Question: [insert decision]
Two ways to run it
Option A — let Claude pick the roles. Use this when you don't know yet what expertise the question needs:
Form a decision council for the question below.
First, select 4–6 distinct advisor roles that are most relevant to this decision.
For each role, state its lens and what it is responsible for evaluating.
Have the advisors respond independently. Require them to identify assumptions,
disagree where warranted, and distinguish facts from inferences.
Then appoint a Chair to synthesize the discussion. The Chair should state:
1. The recommended decision
2. The rationale and key trade-offs
3. The strongest dissenting view
4. The biggest risk or unknown
5. The next three actions, including what must be validated
Decision: [insert question]
Context and constraints: [insert facts, objectives, timing, budget, stakeholders]
Form a decision council for the question below.
First, select 4–6 distinct advisor roles that are most relevant to this decision.
For each role, state its lens and what it is responsible for evaluating.
Have the advisors respond independently. Require them to identify assumptions,
disagree where warranted, and distinguish facts from inferences.
Then appoint a Chair to synthesize the discussion. The Chair should state:
1. The recommended decision
2. The rationale and key trade-offs
3. The strongest dissenting view
4. The biggest risk or unknown
5. The next three actions, including what must be validated
Decision: [insert question]
Context and constraints: [insert facts, objectives, timing, budget, stakeholders]
Form a decision council for the question below.
First, select 4–6 distinct advisor roles that are most relevant to this decision.
For each role, state its lens and what it is responsible for evaluating.
Have the advisors respond independently. Require them to identify assumptions,
disagree where warranted, and distinguish facts from inferences.
Then appoint a Chair to synthesize the discussion. The Chair should state:
1. The recommended decision
2. The rationale and key trade-offs
3. The strongest dissenting view
4. The biggest risk or unknown
5. The next three actions, including what must be validated
Decision: [insert question]
Context and constraints: [insert facts, objectives, timing, budget, stakeholders]
This is the best default for a new or unusual question.
Option B — specify the council yourself. Use this prompt once you know the terrain and want consistent output across repeated analyses.
Run this through a healthcare pricing and revenue council.
Advisors:
1. Pricing strategist — assess willingness to pay, value metric, package design,
price architecture, and competitive positioning.
2. Finance lead — assess revenue impact, margin, forecast variance, downside,
and scenario economics.
3. Customer/renewal lead — assess adoption, procurement friction, renewal risk,
customer equity, and communication implications.
4. Contract and operations lead — assess billing mechanics, contract language,
implementation effort, reporting, and control risks.
5. Skeptic — challenge assumptions, quantify what could fail, and identify
disconfirming evidence.
Process:
- Each advisor responds independently before seeing a synthesis.
- Do not force consensus; surface material disagreements.
- Label facts, assumptions, calculations needed, and recommendations.
Chair:
- Reconcile the views into a clear recommendation.
- State confidence level and why.
- Name the biggest risk and the strongest opposing view.
- Provide the next three actions, the required analyses, and decision criteria.
Decision: [insert decision]
Context: [insert current pricing, customer segments, constraints, data, timeline]
Run this through a healthcare pricing and revenue council.
Advisors:
1. Pricing strategist — assess willingness to pay, value metric, package design,
price architecture, and competitive positioning.
2. Finance lead — assess revenue impact, margin, forecast variance, downside,
and scenario economics.
3. Customer/renewal lead — assess adoption, procurement friction, renewal risk,
customer equity, and communication implications.
4. Contract and operations lead — assess billing mechanics, contract language,
implementation effort, reporting, and control risks.
5. Skeptic — challenge assumptions, quantify what could fail, and identify
disconfirming evidence.
Process:
- Each advisor responds independently before seeing a synthesis.
- Do not force consensus; surface material disagreements.
- Label facts, assumptions, calculations needed, and recommendations.
Chair:
- Reconcile the views into a clear recommendation.
- State confidence level and why.
- Name the biggest risk and the strongest opposing view.
- Provide the next three actions, the required analyses, and decision criteria.
Decision: [insert decision]
Context: [insert current pricing, customer segments, constraints, data, timeline]
Run this through a healthcare pricing and revenue council.
Advisors:
1. Pricing strategist — assess willingness to pay, value metric, package design,
price architecture, and competitive positioning.
2. Finance lead — assess revenue impact, margin, forecast variance, downside,
and scenario economics.
3. Customer/renewal lead — assess adoption, procurement friction, renewal risk,
customer equity, and communication implications.
4. Contract and operations lead — assess billing mechanics, contract language,
implementation effort, reporting, and control risks.
5. Skeptic — challenge assumptions, quantify what could fail, and identify
disconfirming evidence.
Process:
- Each advisor responds independently before seeing a synthesis.
- Do not force consensus; surface material disagreements.
- Label facts, assumptions, calculations needed, and recommendations.
Chair:
- Reconcile the views into a clear recommendation.
- State confidence level and why.
- Name the biggest risk and the strongest opposing view.
- Provide the next three actions, the required analyses, and decision criteria.
Decision: [insert decision]
Context: [insert current pricing, customer segments, constraints, data, timeline]
Do you actually need a chair?
Unless you’re in NYC and waiting in a 500-person line for Froyo, not necessarily. If you skip this step, you get five perspectives you synthesize yourself as the chairman. This is fine for early-stage ideation. But for anything you need to act on — a new hire, an investment call, a stakeholder decision — appoint the chair. Claude writes down what was agreed, what wasn't, and what happens next; it’s a soundboard and decision framework.
Another caveat: don't tell the chair to find "the consensus." Instead, ask for the best-supported recommendation, the strongest dissent, and any conditions that would flip the decision. “Consensus” implies agreement, and that’s counter to the whole operation.
Practical defaults
4-6 advisors. More roles add noise, not coverage.
Make the roles actually different. A "account executive" and a "finance manager" may not agree with each other (it’s exactly the productive friction you're paying for).
Put a skeptic in the room on anything high-stakes.
Feed it real inputs. Actual economics, segment splits, contract constraints, who owns the decision, the timeline, the success metric. The council reasons over what you give it; it doesn't go find the facts itself.
Make it flag what needs validation. A council is a reasoning structure. It is not independent confirmation of anything. Please don’t let a council decide your life decisions.