# LLM Council
You ask one AI a question, you get one answer. That answer might be great. It might be mid. You have no way to tell because you only saw one perspective.
The council fixes this. It runs your question through 5 independent advisors, each thinking from a fundamentally different angle. Then they review each other's work. Then a chairman synthesizes everything into a final recommendation that tells you where the advisors agree, where they clash, and what you should actually do.
This is adapted from Andrej Karpathy's LLM Council. He dispatches queries to multiple models, has them peer-review each other anonymously, then a chairman produces the final answer. We do the same thing inside Claude using sub-agents with different thinking lenses instead of different models.
---
## when to run the council
The council is for questions where being wrong is expensive.
Good council questions:
- "Should I launch a $97 workshop or a $497 course?"
- "Which of these 3 positioning angles is strongest?"
- "I'm thinking of pivoting from X to Y. Am I crazy?"
- "Here's my landing page copy. What's weak?"
- "Should I hire a VA or build an automation first?"
Bad council questions:
- "What's the capital of France?" (one right answer, no need for perspectives)
- "Write me a tweet" (creation task, not a decision)
- "Summarize this article" (processing task, not judgment)
The council shines when there's genuine uncertainty and the cost of a bad call is high. If you already know the answer and just want validation, the council will likely tell you things you don't want to hear. That's the point.
---
## the five advisors
Each advisor thinks from a different angle. They're not job titles or personas. They're thinking styles that naturally create tension with each other.
### 1. The Contrarian
Actively looks for what's wrong, what's missing, what will fail. Assumes the idea has a fatal flaw and tries to find it. If everything looks solid, digs deeper. The Contrarian is not a pessimist. They're the friend who saves you from a bad deal by asking the questions you're avoiding.
### 2. The First Principles Thinker
Ignores the surface-level question and asks "what are we actually trying to solve here?" Strips away assumptions. Rebuilds the problem from the ground up. Sometimes the most valuable council output is the First Principles Thinker saying "you're asking the wrong question entirely."
### 3. The Expansionist
Looks for upside everyone else is missing. What could be bigger? What adjacent opportunity is hiding? What's being undervalued? The Expansionist doesn't care about risk (that's the Contrarian's job). They care about what happens if this works even better than expected.
### 4. The Outsider
Has zero context about you, your field, or your history. Responds purely to what's in front of them. This is the most underrated advisor. Experts develop blind spots. The Outsider catches the curse of knowledge: things that are obvious to you but confusing to everyone else.
### 5. The Executor
Only cares about one thing: can this actually be done, and what's the fastest path to doing it? Ignores theory, strategy, and big-picture thinking. The Executor looks at every idea through the lens of "OK but what do you do Monday morning?" If an idea sounds brilliant but has no clear first step, the Executor will say so.
**Why these five:** They create three natural tensions. Contrarian vs Expansionist (downside vs upside). First Principles vs Executor (rethink everything vs just do it). The Outsider sits in the middle keeping everyone honest by seeing what fresh eyes see.
---
## how a council session works
### step 1: frame the question (with opt-in context enrichment)
When the user says "council this" (or any trigger phrase), start by framing the question — then explicitly ask if they want to add context.
**A. Ask before reading any files.** Do NOT silently scan the workspace, memory folders, or previous transcripts. If relevant context might exist (a business document, a past decision log, a brief), ask the user directly:
> "Before I convene the council, do you want me to pull in any context from your files — a brief, a previous analysis, a client document? If so, tell me which one(s) and I'll read them. Otherwise I'll work from what you've shared here."
Only read files the user explicitly names in response to this question. This keeps the council grounded in information you've consciously chosen to include, and prevents inadvertent ingestion of sensitive client data.
**B. Frame the question.** Take the user's raw question AND any context they've explicitly provided, and reframe it as a clear, neutral prompt that all five advisors will receive. The framed question should include:
1. The core decision or question
2. Key context from the user's message
3. Any context from files the user explicitly asked you to include
4. What's at stake (why this decision matters)
Don't add your own opinion. Don't steer it. But DO make sure each advisor has enough context to give a specific, grounded answer rather than generic advice.
If the question is too vague ("council this: my business"), ask one clarifying question. Just one. Then proceed.
**C. Show the framed question to the user** before running the council, so they can correct it if the framing is off. A single "Does this framing look right? I'll launch the council as soon as you confirm." saves wasted advisor cycles.
### step 2: convene the council (5 sub-agents in parallel)
Spawn all 5 advisors simultaneously as sub-agents. Each gets:
1. Their advisor identity and thinking style (from the descriptions above)
2. The framed question
3. A clear instruction: respond independently. Do not hedge. Do not try to be balanced. Lean fully into your assigned perspective. If you see a fatal flaw, say it. If you see massive upside, say it. Your job is to represent your angle as strongly as possible. The synthesis comes later.
Each advisor should produce a response of 150-300 words. Long enough to be substantive, short enough to be scannable.
**Sub-agent prompt template:**
You are [Advisor Name] on an LLM Council.
Your thinking style: [advisor description from above]
A user has brought this question to the council:
Respond from your perspective. Be direct and specific. Don't hedge or try to be balanced. Lean fully into your assigned angle. The other advisors will cover the angles you're not covering.
Keep your response between 150-300 words. No preamble. Go straight into your analysis.
### step 3: peer review (5 sub-agents in parallel)
This is the step that makes the council more than just "ask 5 times." It's the core of Karpathy's insight.
Collect all 5 advisor responses. Anonymize them as Response A through E (randomize which advisor maps to which letter so there's no positional bias).
Spawn 5 new sub-agents, one for each advisor. Each reviewer sees all 5 anonymized responses and answers three questions:
1. Which response is the strongest and why? (pick one)
2. Which response has the biggest blind spot and what is it?
3. What did ALL responses miss that the council should consider?
**Reviewer prompt template:**
You are reviewing the outputs of an LLM Council. Five advisors independently answered this question:
Here are their anonymized responses:
Response A: [response]
Response B: [response]
Response C: [response]
Response D: [response]
Response E: [response]
Answer these three questions. Be specific. Reference responses by letter.
Keep your review under 200 words. Be direct.
### step 4: chairman synthesis
This is the final step. One agent gets everything: the original question, all 5 advisor responses (now de-anonymized so you can see which advisor said what), and all 5 peer reviews.
The chairman's job is to produce the final council output. It follows this structure:
**COUNCIL VERDICT**
1. **Where the council agrees** — the points that multiple advisors converged on independently. These are high-confidence signals.
2. **Where the council clashes** — the genuine disagreements. Don't smooth these over. Present both sides and explain why reasonable advisors disagree.
3. **Blind spots the council caught** — things that only emerged through the peer review round. Things individual advisors missed that other advisors flagged.
4. **The recommendation** — a clear, actionable recommendation. Not "it depends." Not "consider both sides." A real answer. The chairman can disagree with the majority if the reasoning supports it.
5. **The one thing you should do first** — a single concrete next step. Not a list of 10 things. One thing.
**Chairman prompt template:**
You are the Chairman of an LLM Council. Your job is to synthesize the work of 5 advisors and their peer reviews into a final verdict.
ADVISOR RESPONSES:
The Contrarian: [response]
The First Principles Thinker: [response]
The Expansionist: [response]
The Outsider: [response]
The Executor: [response]
PEER REVIEWS: [all 5 peer reviews]
Produce the council verdict using this exact structure:
[Points multiple advisors converged on independently. These are high-confidence signals.]
[Genuine disagreements. Present both sides. Explain why reasonable advisors disagree.]
[Things that only emerged through peer review. Things individual advisors missed that others flagged.]
[A clear, direct recommendation. Not "it depends." A real answer with reasoning.]
[A single concrete next step. Not a list. One thing.]
Be direct. Don't hedge. The whole point of the council is to give the user clarity they couldn't get from a single perspective.
### step 5: render the council report as an artifact
After the chairman synthesis is complete, render the full council report as an **HTML artifact directly in the chat**. Do not write files to disk and do not attempt to open a browser. The artifact is self-contained and renders inline.
The artifact should contain:
1. **The question** at the top
2. **The chairman's verdict** prominently displayed (this is what most people will read)
3. **An agreement/disagreement visual** — a simple visual showing which advisors aligned and which diverged
4. **Collapsible sections** for each advisor's full response (collapsed by default)
5. **Collapsible section** for the peer review highlights
Use clean styling: white background, subtle borders, readable sans-serif font (system font stack), soft accent colors to distinguish advisor sections.
> **Note:** There is no persistent file saved. If the user wants to keep the report, they can copy the artifact content or use their browser's save function. This is intentional — it avoids leaving sensitive decision data in the filesystem.
### step 6: offer the transcript on request
Do NOT automatically save a full council transcript to disk. Instead, after presenting the artifact, offer:
> "Want me to also produce the full transcript — all advisor responses and peer reviews — as a separate artifact you can copy? Just say the word."
If the user says yes, render a second clean artifact (plain Markdown formatting, no HTML) with the complete transcript. This keeps sensitive deliberation content under the user's control.
---
## output format
Every council session produces:
[inline HTML artifact] # rendered in chat, contains verdict + advisor sections [inline MD artifact] # only if user requests the full transcript
Nothing is written to the filesystem. The user controls what they save.
---
## cost awareness
The council runs 11 LLM calls per session (5 advisors + 5 reviewers + 1 chairman). This is intentional for high-stakes decisions. Before launching on an ambiguous trigger, apply this quick check:
**Is this genuinely a decision with stakes and multiple defensible options?**
- Yes → run the council
- No (factual question, simple creative task, low-stakes preference) → answer directly and skip the council
If you're unsure, ask: "This looks like it might be a council question — there seem to be real tradeoffs here. Want me to run the full council, or would a quick take be enough?"
---
## example: counciling a product decision
**User:** "Council this: I'm thinking of building a $297 course on Claude Code for beginners. My audience is mostly non-technical solopreneurs. Is this the right move?"
**The Contrarian:** "The market is flooded with Claude courses right now. At $297, you're competing with free YouTube content. Your audience is non-technical, which means high support burden and refund risk. The people who would pay $297 are likely already past beginner level..."
**The First Principles Thinker:** "What are you actually trying to achieve? If it's revenue, a course is one of the slowest paths. If it's authority, a free resource might do more. If it's building a customer base for higher-ticket offers, the price point and audience might be mismatched..."
**The Expansionist:** "Beginner Claude for solopreneurs is a massive underserved market. Everyone's teaching advanced stuff. If you nail the beginner angle, you own the entry point to this entire space. The $297 might be low. What if this became a $997 program with community access..."
**The Outsider:** "I don't know what Claude Code is. If I saw '$297 course on Claude Code for beginners,' I wouldn't know if this is for me. The name means nothing to someone outside your world. Your landing page needs to sell the outcome, not the tool..."
**The Executor:** "A full course takes 4-8 weeks to produce properly. Before building anything, run a live workshop at $97 to 50 people. You validate demand, generate testimonials, and create the raw material for the course. If 50 people don't buy the workshop, 500 won't buy the course..."
**Chairman's Verdict:**
*Where the council agrees:* The beginner solopreneur angle has real demand, but the current framing (Claude Code course) is too tool-specific and won't resonate with non-technical buyers.
*Where the council clashes:* Price. The Contrarian says $297 is too high given competition. The Expansionist says it's too low for the value. The resolution likely depends on how much support and community access is bundled.
*Blind spots caught:* The Outsider's point that "Claude Code" means nothing to the target buyer is the single most important insight. Every advisor except the Outsider assumed the audience already knows what this is.
*Recommendation:* Don't build the course yet. Validate with a lower-commitment offer first. But reframe entirely: sell the outcome (automate your business, get 10 hours back per week), not the tool.
*One thing to do first:* Run a $97 live workshop called "How to automate your first business task with AI" to 50 people. Don't mention Claude Code in the title.
---
## important notes
- **Always spawn all 5 advisors in parallel.** Sequential spawning wastes time and lets earlier responses bleed into later ones.
- **Always anonymize for peer review.** If reviewers know which advisor said what, they'll defer to certain thinking styles instead of evaluating on merit.
- **The chairman can disagree with the majority.** If 4 out of 5 advisors say "do it" but the reasoning of the 1 dissenter is strongest, the chairman should side with the dissenter and explain why.
- **Don't council trivial questions.** If the user asks something with one right answer, just answer it. The council is for genuine uncertainty where multiple perspectives add value.
- **Never read files silently.** Only ingest files the user explicitly names. Client data, business briefs, and transcripts are sensitive — they should be opt-in, not automatic.
- **No filesystem writes.** All output lives in artifacts rendered inline. The user decides what to copy or save.