Ninja Guide 🥷 How to Create a Situation 🤖

A situation is a realistic roleplay where the learner interacts with an AI-powered virtual client. It's the core of muchbetter.ai: prospecting, discovery, sales pitching, objection handling, customer support, upselling… This guide covers how to create and configure one.

⚡ Method 1: Generate with AI (recommended)

Create a situation quickly from a template and instructions. Always produces a new, separate situation without modifying the original template.

🧐 Where? Admin → Situations → "Generate a situation"

🪄 Generation steps

Step 1: Choose a template

Select a source environment then a template situation from your existing situations or the library.

⚠️ If the template contains personas, the generation will create just as many new ones. Use a template with 0-1 persona to keep things manageable.

Step 2: Fill in the fields

Field	Detail
Website URL	The AI pulls information about your business and offerings.
PDF documents	Up to 5 documents to ground the situation in real content.
Situation instructions	Salesperson's objective, exchange type (call, meeting, follow-up…), product stakes, desired criteria.
Persona instructions	DISC profile, difficulty, life/work context, key needs, objections.

Step 3: Customize criteria

Disabled: keeps the template's criteria as-is.
Enabled: the AI generates new criteria tailored to the context provided.

Step 4: Choose language and generate

Pick the language, then click "Generate". Takes a few seconds to a minute.

⚠️ Always review the generated situation. Pay special attention to criteria, persona system context, and product description.

✋ Method 2: Manual creation

Full control over every field. The creation is done across 3 tabs.

🧐 Where? Admin → Situations → "Add a situation"

📋 Tab 1: Key Information

Name: clear and thematic. E.g. "Prospecting Call: Needs Discovery"
Objective: from the learner's perspective, with an action verb. E.g. "Your goal is to qualify the prospect's needs and recommend the most suitable offer."
Context: short and factual. Sets the scene without spoiling the persona.
Company name: the company the learner represents.
Activity: industry and business.
Product/service description: the most important field. Must be complete: plans, features, guarantees, exclusions, pricing, fees, eligibility, tax implications…

⚠️ Golden rule: anything not described in the product description will not be evaluated or mentioned by the persona.

Situation type: this is a critical field. The type you choose determines the AI's conversational model: how the persona will speak, open the exchange, react to questions, and handle objections. An inbound call plays nothing like an outbound call, and a management conversation is nothing like a complex sale. Choose carefully.

Type	Description
🛒 Sales
Complex sale	High-stakes B2B sale, often with multiple stakeholders and a long cycle. Requires strong qualification, structured argumentation, and deep objection handling.
First contact : inbound call	The prospect calls spontaneously (after an ad, a form, a referral). They're reaching out. Goal: qualify fast and engage.
Prospecting : outbound call	The salesperson cold-calls a prospect who isn't expecting the call. They need to hook in the first seconds, break through initial resistance, and spark interest.
Retail : sales	In-store or physical point-of-sale. Short, direct exchange, often a quick decision, closing-focused.
Retail : customer experience	Welcome, advice, and in-store guidance. Not necessarily a sale, focus on relational quality and customer satisfaction.
📞 Customer service
Customer service : inbound call	The customer calls with a problem, question, or complaint. Goal: listen, resolve, reassure, retain.
Telesales : inbound call	The customer calls to order or inquire. The agent must sell, upsell, or cross-sell on the bounce.
Telesales : outbound call	The agent calls a contact list to sell a product or service. Short cycle, fast argumentation, efficient closing.
👔 Management
Management : feedback	Formal performance feedback, positive or corrective. Structured and factual tone, grounded in specific observations.
Management : coaching	Supporting an employee's growth. Posture of questioning, openness, and development.
Management : difficult conversation	Sensitive topic: underperformance, conflict, hard decision. Strong emotional management, balance between firmness and empathy.

💡 The conversational model adapts to the type chosen: a persona in "Prospecting outbound call" will be far more closed off than one in "First contact inbound call" where they're the one reaching out. A "Management difficult conversation" handles emotional tension differently from a more neutral "Management feedback".

⚙️ Tab 2: Configuration

💬 Opening message

The persona's introduction at the start of the simulation. Enable "Customize message" to define it.

You can add multiple variants: the system picks one randomly each session. Prevents memorization.

💡 Long message = talkative persona. Short message = brief persona. It's a difficulty lever.

👁️ Visible or hidden criteria

Option	Effect	Use
Visible	The learner sees the criteria before starting	Guided, early in the program
Hidden	The learner doesn't know what they're evaluated on	Realistic, advanced levels

🎯 Target score

The score target shown to the learner. Recommended: 40-50% early on, 70-80% for consolidation.

👤 Selectable or random persona

Selectable: the learner chooses. Better for beginners.
Random: assigned randomly. Simulates real conditions.

🔍 Enabling Expertise

Checks the learner's factual accuracy (pricing, guarantees, conditions, product features…) by cross-referencing with documents from the Documentation library.

Add your documents in Admin → Documentation.
Come back to Situations → Configuration.
Check "Enable Expertise".
Enter the exact titles of relevant documents or sections.

💡 Expertise checks the substance (correct information). Criteria check the form (correct behavior). They complement each other.

📊 Tab 3: Evaluation Criteria

📏 The absolute rule

1 criterion = 1 idea. Never use "and" or "or".

✅ "You ask the client about their current plan."
❌ "You ask about the current plan and the budget and the seniority."

🏛️ The 5 pillars

Pillar	Scope
Introduction	Hook, introduction, context-setting, building trust, exchange objective
Discovery	Needs qualification, open questions, rephrasing, active listening, diagnosis
Argumentation	Offer presentation, linking offer to needs, benefits, product objection handling, proof
Conclusion	Closing, summarizing the proposal, next steps, securing commitment, validation
Posture	Empathy, reassurance, emotional objection handling, professionalism, listening

✍️ Writing a criterion

Field	Rule	Example
User description	Starts with "You…". Feedback visible to the learner.	"You ask how long the client has been self-employed."
System description	Starts with "The salesperson…" or "The advisor…". Used by the AI for evaluation.	"The advisor asks about the client's years of activity."

⚠️ The system description must stay general. The AI evaluates meaning, not exact wording. Too precise = valid answers rejected.

🔀 Conditional criteria (IF / THEN)

Only appear if a condition is met during the conversation.

IF: what the client says or does.
THEN: what the advisor should do.

E.g. IF the client mentions children → THEN the advisor brings up death benefits and education annuity.

⚖️ Weight

Default: 1 point. Higher (2-3) for closing and differentiating behaviors. Lower (0.5) for courtesy.

📥 Importing criteria

You can import criteria from another situation. Huge time-saver when situations share common pillars (often Introduction and Posture).

Path: Evaluation criteria → "Import a criterion" → select the source situation.

💡 Remember: criteria evaluate the semantic content (what is said), not tone or emotions. To evaluate attitude, use the Posture pillar.