When should you use Multi-Persona Debate / Council?

Decisions with multiple legitimate perspectives. Strategic or ethical questions where bias-checking matters. Brainstorming where you want range, not depth. Pre-commitment audits of plans and pitches.

When NOT to use Multi-Persona Debate / Council?

Single-answer factual questions. Tasks where one expert clearly dominates the domain. When you need fast output — debates are token-hungry.

How does Multi-Persona Debate / Council work?

Define 3–5 personas with distinct cognitive biases (e.g. Skeptic, Pragmatist, Visionary, Cynic, Outsider). Round 1: each persona gives a brief opening position. Round 2: cross-examination — each persona challenges another. Round 3: rebuttals, with required concessions. A moderator (still the same model) synthesizes: the shared unstated assumption, the sharpest disagreements, and an integrated recommendation.

All techniques

Glossary · Technique

Multi-Persona Debate / Council

Also known as: Multi-agent debate, Expert panel

Simulate multiple distinct experts debating the question, then synthesize. Surfaces what a single persona would have missed.

Try the interactive template

When to use it

Decisions with multiple legitimate perspectives.
Strategic or ethical questions where bias-checking matters.
Brainstorming where you want range, not depth.
Pre-commitment audits of plans and pitches.

When not to use it

Single-answer factual questions.
Tasks where one expert clearly dominates the domain.
When you need fast output — debates are token-hungry.

How it works

1Define 3–5 personas with distinct cognitive biases (e.g. Skeptic, Pragmatist, Visionary, Cynic, Outsider).
2Round 1: each persona gives a brief opening position.
3Round 2: cross-examination — each persona challenges another.
4Round 3: rebuttals, with required concessions.
5A moderator (still the same model) synthesizes: the shared unstated assumption, the sharpest disagreements, and an integrated recommendation.

Example

Lazy prompt

What do you think of our new pricing?

Using the technique

Convene 5 personas (Skeptic, CFO, Customer, Sales Lead, Outsider) to debate this pricing. Three rounds, each must concede a point. End with the assumption all 5 implicitly shared.

Common pitfalls

If personas are too similar, the debate is theater — output is a wash.
Forcing concessions is essential or the model defends all sides equally.
Without naming the shared assumption at the end, you miss the most valuable signal.

Where this came from

Liang et al., 2023 — "Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debate."

Related techniques

Tree-of-Thoughts (ToT) Prompting

Generate multiple reasoning branches per step, evaluate each, and prune. Beats single-path Chain-of-Thought on hard decisions.

Adversarial / Red-Team Prompting

Ask the model to attack your idea before defending it. Surfaces the failure modes before they ship.

Self-Refine

Generate → critique own output → revise → repeat. Pushes a model's output much closer to its capability ceiling.

Try it interactively

The interactive template lets you fill in your scenario and generates a copy-ready prompt that uses this technique.

Open the template