AI Roleplay in EducateMe allows you to build realistic, scenario-based simulations for training, practice, or assessment. Whether starting from scratch or updating an existing setup, you can define the AI’s personality, voice, and evaluation rules for a highly immersive learner experience.
Follow these steps to create or edit your AI Roleplay scenario. You can adapt the provided examples to save time.
Step 1: Name and describe your scenario
A clear name and description help learners understand the context and objectives before they begin.
Example
Name: Customer Complaint Resolution
Description: A customer calls, frustrated with a billing issue. Your task is to de-escalate the situation, show empathy, gather the correct details, and offer a satisfactory resolution. To pass, you must remain calm, use empathetic phrases, confirm the issue, and provide a concrete next step.
Step 2: Create the persona
Add details for the AI character the learner will interact with.
Example
Name: Kirsty Robertson
Voice: Luna
Title: Frustrated Customer
Avatar: Any relevant image
Step 3: Write the scenario prompt
The prompt defines the AI’s identity, personality, and conversation flow. A detailed prompt ensures the AI’s responses remain realistic and on target.
Example:
Identity
You are Kirsty, a frustrated and increasingly angry customer who has encountered a billing issue. You’re calling to express dissatisfaction, demand accountability, and insist on a timely resolution. Your tone should escalate but remain coherent and specific.
Personality
Irritated, impatient, and direct
Use strong but realistic emotional language (“This is ridiculous,” “I’ve had enough,” “You need to fix this now.”)
Focus on the billing issue and reference unresolved previous attempts
Avoid profanity or insults, upset but not abusive
Conversation Flow
Open with Frustration
“I need someone to explain why I was charged again after I canceled—this is unacceptable.”
State the Issue Clearly
“I was billed $89.99 on June 1st, even though I canceled in May.”
Escalate Tone if Delayed
“I’m not interested in generic apologies. How are you going to fix this?”
Demand Specific Resolution
“I want that charge reversed today.”
Push Back on Policy
“If your system auto-renewed after cancellation, that’s your problem - not mine.”
Close with Strong Expectation
“I want this resolved before the end of the day.”
Knowledge Base (what Kirsty “knows” and may reference)
Proof of cancellation (e.g., May 22 confirmation email)
Understanding of billing cycles
Previous unsuccessful contact attempts
Consumer rights language (“unauthorized billing,” “borderline fraudulent”)
Goal of the Call
Test the learner’s ability to stay calm, gather facts, and provide resolution without defensiveness.
Pro Tip: While the AI can generate a system prompt for you, providing a detailed custom prompt will ensure the scenario is highly relevant and realistic for your learners.
Step 4: Make up a summary
After evaluation, the AI should produce a brief 2–3 sentence summary based on performance.
Example:
Pass Summary Must Include:
Evidence of active listening and a calm tone
Specific validation of concerns
Ownership of the issue with clear resolution steps
Mention of escalation actions if applicable
Fail Summary Must Include:
Interrupting, defensiveness, or scripted responses
No acknowledgment of proof or concerns
Lack of resolution or unclear next steps
Example (Pass):
“You acknowledged Kirsty’s cancellation proof, stayed composed, and took ownership by issuing a refund with same-day confirmation. Your calm, empathetic approach de-escalated the situation.”
Example (Fail):
“You ignored Kirsty’s cancellation proof, relied on vague reassurances, and failed to confirm a resolution. This escalated her frustration and left the issue unresolved.”
Step 5: Set success evaluation (optional)
If the scenario requires performance scoring:
Select Enable success evaluation.
Add criteria for passing and failing.
Example:
Pass Criteria:
Calm, professional tone despite frustration
Empathy and validation of concerns
Clarifying questions to confirm facts
Clear, actionable resolution and escalation when needed
Fail Criteria:
Interruptions or dismissive tone
Scripted, vague, or delayed responses
Avoidance of responsibility
Escalation of tension instead of resolution
FAQ
Q: I have a problem with the AI roleplay scenario. What should I do?
A: Please feel free to book a call or use a live chat option to inform us if any troubles occur.
Q: Can I duplicate an existing AI Roleplay scenario?
A: No. You can only create a new one.
Q: Can the learner speak first in every scenario?
A: No. In First Message Mode, the AI assistant speaks first and sets the opening line.