Can Personas Talk to Each Other? How to Set It Up?

Question Type

User Manual Question


Core Answer

Yes. This is the Discussion Chat (Group Discussion) feature.

AI personas can dialogue with each other, exchange viewpoints, reach consensus or diverge.


Two Modes Compared

Interview Chat (One-on-One Interviews)

Dialogue Structure:

  • Interviewer ↔ AI Persona
  • AI personas do NOT talk to each other
  • Each interview proceeds independently

Diagram:


Discussion Chat (Group Discussion)

Dialogue Structure:

  • AI moderator guides discussion
  • AI personas talk to each other
  • Viewpoints collide, reach consensus or divergence

Diagram:


How to Set Up Discussion Chat?

Step 1: Create Discussion Project

Entry Point: Homepage → "Create New Interview" → Select "Discussion Chat"

Fill in Information:

  • Project Name: e.g., "Sparkling Coffee Pricing Discussion"
  • Research Objective: Brief description
  • Discussion Type:
    • Focus Group
    • Debate
    • Roundtable

Step 2: Select AI Personas (3-8 People)

Recommended Quantity:

  • Standard: 5-6 people
  • Minimum: 3 people (insufficient viewpoint collision)
  • Maximum: 8 people (beyond 8, discussion hard to control)

Persona Mix Recommendations:

  • ✅ Different stances: price-sensitive vs quality-first
  • ✅ Different backgrounds: beginners vs advanced users
  • ✅ Opposing viewpoints: supporters vs opponents

Example: Sparkling coffee pricing discussion (5 people)

  • 2 price-sensitive types (focus on value)
  • 2 health-conscious types (focus on ingredients)
  • 1 social trendsetter type (focus on appearance)

Step 3: Set Discussion Topic

Topic Types:

  • Comparison validation: Option A vs Option B
  • Viewpoint collision: Different users' perspectives
  • Consensus exploration: Common ground
  • Controversy identification: Most controversial points

Example Topics:

  • "Do you think ¥18 sparkling coffee is expensive?"
  • "Subscription vs one-time purchase, which is better?"
  • "Can AI personal trainer replace human coach?"

Step 4: Choose Discussion Type

Focus Group

Characteristics:

  • AI moderator guides discussion
  • Free expression of viewpoints
  • Seek consensus and divergence

Use For:

  • Product concept testing
  • Feature acceptance
  • User attitude research

Debate

Characteristics:

  • Clear pro and con sides
  • Opposing viewpoint collision
  • In-depth argumentation

Use For:

  • Solution comparison (A vs B)
  • Controversial topic exploration
  • Decision support

Example:

  • Pro: Support subscription (3 people)
  • Con: Support one-time purchase (3 people)

Roundtable

Characteristics:

  • Equal speaking rights
  • No clear stance
  • Open-ended exploration

Use For:

  • Exploratory research
  • Brainstorming
  • Multi-angle insights

Discussion Chat Core Capabilities

Capability 1: AI Moderator Guidance

Moderator Responsibilities:

  • Control speaking pace
  • Ensure everyone has opportunity to express
  • Identify viewpoint conflicts, actively guide in-depth discussion
  • Ask "Why do you view this differently?"

Example:


Capability 2: Viewpoint Collision and Evolution

Observe:

  • Initial viewpoint vs post-discussion viewpoint
  • Who convinced whom?
  • Which viewpoints gained resonance?

Example:


Capability 3: Identify Consensus and Divergence

Auto-Identify:

  • Consensus Areas: Points everyone agrees on
  • Divergence Areas: Opposing opinions
  • Controversy Points: Most heated discussion topics

Output:

  • Consensus list: "0-sugar concept is attractive"
  • Divergence list: "Price acceptance varies greatly"
  • Controversy point: "Is sparkling + coffee innovation or gimmick?"

Real-World Cases

Case 1: SaaS Pricing Discussion (6 People)

Research Objective: Monthly vs annual payment, which do users prefer?

Persona Mix:

  • 3 small business decision-makers (prefer monthly)
  • 3 mid-size business decision-makers (prefer annual)

Discussion Process:

Moderator: "Which do you think is better, monthly or annual payment?"

Small Business A: "I prefer monthly, flexible, don't want to pay too much at once" Mid-Size Business B: "We prefer annual, has discount, and budget is annual planning"

Moderator: "Why do small businesses focus on flexibility, mid-size on discounts?"

Small Business A: "Our cash flow is tight, monthly payment doesn't pressure finances" Mid-Size Business B: "We have budget planning, annual discount more cost-effective, reduces procurement frequency"

Results:

  • Consensus: Powerful features are core value
  • Divergence: Payment method preference varies greatly
  • Insight: Company size determines payment preference

Actions:

  • Offer both monthly and annual payment options
  • 20% discount for annual payment
  • Emphasize flexibility to small businesses, ROI to mid-size businesses

Case 2: Fitness App Feature Discussion (5 People)

Research Objective: AI personal trainer feature acceptance

Persona Mix:

  • 2 fitness beginners (support AI)
  • 2 advanced fitness enthusiasts (doubt AI)
  • 1 neutral observer

Discussion Process:

Moderator: "Do you think AI personal trainer can replace human coach?"

Beginner A: "I think so, AI won't despise my incorrect form" Advanced B: "How can AI know my body condition? My knee is injured, does AI know?"

Moderator: "What if AI lets you input injury information?"

Advanced B: "Then might consider, but still worry AI lacks professionalism" Beginner A: "For me AI is enough, human coach too expensive"

Results:

  • Consensus: Form correction is essential need
  • Divergence: Beginners trust AI, advanced users doubt AI professionalism
  • Insight: Users don't reject AI, but worry "AI doesn't understand me"

Actions:

  • Add "injury input" feature
  • Promote AI trainer to beginners
  • Promote "AI assist + human guidance" combo to advanced users

Common Questions

Q1: What's the difference between Discussion and Interview?

DimensionInterviewDiscussion
Dialogue StyleInterviewer ↔ AI PersonaAI Persona ↔ AI Persona
Follow-up Depth★★★★★★★★☆☆
Viewpoint Collision❌ None✅ Yes
Consensus Identification★★★☆☆ (manual integration)★★★★★ (auto-identify)
Use CasesExplore "why"Compare views, quick consensus

Q2: Can I insert questions mid-discussion?

Yes.

  • During discussion, you can insert new questions anytime
  • AI moderator will naturally guide to new topic
  • Won't affect discussion coherence

Q3: How to ensure people with different stances all speak?

AI moderator actively controls:

  • Actively call on speakers: "XX, what's your view?"
  • Invite "silent ones" to express
  • Control "chatterers," ensure balanced speaking

Q4: Can Discussion duration be controlled?

Yes. You can set:

  • Minimum duration: Ensure sufficient discussion
  • Maximum duration: Avoid lengthy discussion
  • Or let AI moderator auto-judge (end after viewpoints fully collide)

Best Practices

Recommendation 1: Interview First, Then Discussion

Best Process:

  1. First use Interview to deeply explore individual motivations
  2. Discover key controversy points
  3. Then use Discussion to observe viewpoint collisions

Advantages:

  • First understand individuals, then observe groups
  • Discussion topics more precise

Recommendation 2: Choose Opposing Viewpoint Personas

Don't:

  • ❌ 5 people with similar viewpoints (no collision)

Do:

  • ✅ 2-3 supporters + 2-3 opponents
  • ✅ Ensure viewpoint diversity

Recommendation 3: Use Debate Mode to Test Solutions

Scenario: A/B solution comparison

  • Pro: 3 people support Solution A
  • Con: 3 people support Solution B
  • Observe which side argues more fully

Output:

  • Discover advantages/disadvantages of Solution A/B
  • Users' real preferences

Final Takeaway

"Discussion lets AI personas dialogue and collide, Quickly discover consensus and divergence, not replacing Interview but complementing it."

Remember:

  • ✅ Discussion = AI personas talk to each other
  • ✅ Interview = Interviewer talks to AI personas
  • ✅ Best practice: Interview deep dive first + Discussion viewpoint collision later

Related Feature: Interview vs Discussion Document Version: v2.1

Last updated: 2/9/2026