Skip to main content
Learn how to create effective AI roleplay scenarios from start to finish - from picking your setup method to publishing and refining
Building a great roleplay doesn’t require perfection on the first try. Use the guide below to go deeper on any section covered in the video.

​
Before You Build: The Four Pillars

Before diving into any building method, think through these four elements. They apply regardless of how you create your scenario.
Define the basics of the situation:
  • What role does the learner play? (SDR, AE, Manager, etc.)
  • Who is the AI character?
  • What type of conversation is this? (Cold call, scheduled meeting, video call)
  • What company or product is the learner representing?
Example: “I am a sales representative at Exec. I’m calling a Director of Sales Enablement at a mid-market tech company.”
Be specific about what success looks like. A clear goal keeps the scenario focused.
  • Book a meeting with an AE
  • Gather discovery information
  • Deliver difficult feedback
  • Close a deal
  • Get a clear action item
Example: “My goal is to schedule a discovery meeting and identify key pain points during this initial conversation.”
If you were watching the conversation, what would tell you it’s going well? Define observable success criteria.
  • Cold calls: Booking the meeting
  • Discovery: Gathering specific info (budget, timeline, pain points)
  • Feedback: Getting commitment to behavior change
  • Negotiation: Reaching specific terms
Example: “Success looks like booking the meeting AND uncovering at least 2-3 pain points.”
What makes this conversation difficult? This is what creates realistic practice.
  • Skeptical or resistant prospect
  • Bad timing or previous negative experience
  • Competitor already in play
  • Budget constraints
  • Difficult personality type
Example: “The prospect filled out a form wanting a free trial, but got a sales call instead. They’re frustrated and just want an email.”

​
Choosing a Building Method

Demo Exec Com Roleplays(3)
There are four ways to build a scenario in the Scenario Studio. Choose based on your situation.
MethodBest ForControl Level
Voice SetupBeginners, brainstormingMedium
Chat / TextQuick builds, clear visionMedium–High
TemplatesStructured approach, scalingHigh
Templates + ResourcesScale, consistency, non-SMEsHighest

​
Voice Setup

Create a scenario in 3–5 minutes through a natural voice conversation with the AI Agent. Best for: Beginners, quick creation, people who prefer talking over typing.
1

Navigate to Scenario Studio

From your dashboard, open the Scenario Studio.
2

Click Voice Setup

Select Voice Setup to start a voice conversation with the AI Agent.
3

Describe your scenario

Talk through what you want to build. The agent will ask clarifying questions to fill in any gaps. Keep the four pillars in mind as you speak — you don’t need to be formal.
4

Review and publish

Once the agent has enough detail, it will generate a draft. Review it and publish when you’re ready.
Screenshot2026 02 17at3 07 22PM
You can’t review the voice conversation after ending the call, and you won’t be able to attach documents until after the call is complete. If you need to attach materials upfront, use Chat or Templates instead.
Sample conversation: “I want to practice a follow-up call with a prospect who filled out a form on our website. They’re a Sales Enablement Manager at a mid-market tech company. They wanted a free trial but got scheduled for a call instead, so they’re a bit frustrated. I want to practice booking a meeting while doing some light discovery…”

​
Chat / Text Input

Type your scenario details directly into the chat. Best for: When you have a clear vision and want to move quickly, or when you’re pasting in a prewritten outline.
1

Navigate to Scenario Studio

From your dashboard, open the Scenario Studio.
2

Type your scenario details

Enter your scenario directly into the chat. You can attach documents from the start to enhance the creation process.
3

Answer any follow-up questions

The AI Agent will ask clarifying questions if it needs more detail. The more specific you are upfront, the fewer follow-ups you’ll get.
4

Review and refine

Review the generated draft and adjust through chat as needed.
Sample prompt structure: 
I am a [JOB TITLE] at [COMPANY], and I want to build a [CONVERSATION TYPE] scenario where I am speaking with a [PROSPECT/CHARACTER TITLE] at a [COMPANY TYPE/INDUSTRY].

My goal for this conversation is to [SPECIFIC OUTCOME].

For me to succeed, I must [DEFINITION OF GOOD].

This conversation is challenging because [KEY CHALLENGE].
You can be brief or detailed — the agent will ask follow-up questions for anything it needs. Paste in prewritten outlines to speed things up even more.

​
Templates

Select a conversation template and fill out the form fields. Best for: Structured approach, ensuring you don’t miss key elements, scaling to multiple scenarios. Templates help the AI Agent understand the type of conversation you’re building. Each conversation type has nuances — discovery is about digging deeper, objection handling has different dynamics than demos or feedback conversations. Templates encode those nuances so the agent builds appropriately.
1

Navigate to Scenario Studio

From your dashboard, open the Scenario Studio.
2

Click Use Template

Hover over the conversation type (Sales, Management, HR, etc.) and select a specific template.
3

Fill out the form fields

Complete the fields provided — company context, customer profile, situation details, desired outcomes, and competitive considerations.
4

Submit and review

Submit the form and review the generated draft. Refine through chat as needed.
Demo Exec Com Chat New
Templates take more time to fill out than voice or chat, but they produce the most consistent results — especially when building at scale.

​
Templates + External Resources (Power User Method)

Attach your existing enablement materials to a template and let the AI Agent do the heavy lifting. Best for: Leveraging existing resources, building at scale, scenarios where you’re not the subject matter expert.
1

Select a template

Choose the template that matches your conversation type.
2

Attach your resources

Upload relevant documents — enablement guides, call transcripts, competitive intel, product docs, previous scenario outlines.
3

Add a prompt

At the top of the chat, tell the agent to use your materials to fill out the template. Example:
I am an Account Executive at [COMPANY].
I'll be talking with a [PROSPECT TITLE] at a [COMPANY TYPE].
Please use the attached resources to fill out this template, then provide me with the completed version.

[PASTE TEMPLATE BELOW]
4

Review the completed template

Before building, verify the agent pulled in the right information. Check for accuracy and refine if needed (e.g., “Use the MEDDIC framework for the evaluation criteria”).
5

Build the scenario

When the template looks right, say “let’s build” to generate the draft.
Demo Exec Com Chat New(1)
Not everything from attachments automatically gets pulled into the final scenario. If specific information is missing, explicitly tell the agent to include it.

​
Context vs. Attachments

ContextAttachments
ReusabilityReusable across multiple scenariosSpecific to one scenario
PersistenceSaved for future useNot saved after that session
Best forCommon reference materials (pricing guides, company info)Scenario-specific documents
Example: Save a pricing document that applies to all negotiation scenarios in Context. Use Attachments for a specific call transcript you’re referencing for one build.

​
Testing Before Publishing

Always test your scenario before publishing.
1

Click Try Now

In the Scenario Studio, click Try Now to start a short test conversation with the character.
2

Test specific triggers

Try the moments that matter most — how the character handles resistance, how much info they volunteer, whether the opening line feels natural.
3

Return to chat to adjust

If something feels off, go back to chat and be specific:
  • “Make the character more skeptical”
  • “The character is giving away info too quickly”
  • “Add more resistance around budget discussions”
4

Publish when ready

Click Publish once the scenario feels right.
Demo Exec Com Chat New(2)
Before publishing, review the Conversation Guidelines section carefully. These are the triggers and responses that define how the character behaves — and they can’t be edited after publishing without remixing the scenario.

​
Refining After Publishing

Once a scenario is live, you have two paths for making changes.

​
Edit Directly (No Remix Needed)

These elements can be updated without returning to the studio:
  • Scenario context and objectives
  • Evaluation criteria (Good/Fair/Poor definitions, skills, ordering)
  • Basic details (screen sharing, phone ring)
  • Character details: name, title, voice, backstory, personality traits

​
Edit in Scenario Studio (Remix Required)

These elements require remixing:
  • Response guidelines (how the character behaves in conversation)
  • Conversation flow rules and triggers
  • Hidden details the character knows but doesn’t volunteer
  • Response length and natural speech patterns
Remix creates a copy — your original scenario stays intact with all its session data preserved.
Common remix requests:
  • “Make a harder version of this scenario”
  • “Make the character more resistant to sharing information”
  • “Add a competitor already in play”
  • “Remove budget authority from this character”
Demo Exec Com Chat New(3)

​
Advanced Tuning

Once you have a working scenario, use these techniques to fine-tune character behavior. Natural speech patterns 
Be sure that the character uses occasional filler words such as "hmm," "well," "um," and natural pauses.
Controlling response length 
Keep responses to 1-2 sentences maximum unless asked to elaborate. This is a busy manager on a call, not someone who monologues or overshares.
Adding discovery depth 
On some of these response guidelines, add one or two more levels of depth to make discovery more difficult.
Multi-level responses work like this:
  • Level 1 (Initial): Surface-level, guarded response
  • Level 2 (Follow-up): More detail if pressed
  • Level 3 (Final): Full information only with strong probing
Using transcripts to improve scenarios After running through a full session, copy the transcript and paste it into a remix. Tell the agent what didn’t feel right — this is much more effective than trying to describe issues from memory.

​
Customizing Evaluation Criteria

Evaluation criteria determine how learners are scored. You can customize these at any time after publishing. What you can do:
  • Add, remove, or reorder criteria
  • Edit the Good/Fair/Poor descriptions
  • Add scenario-specific context within each description
  • Associate skills with each criterion for analytics tracking
  • Import a custom rubric by clearing existing criteria and adding your own
Criteria count by difficulty:
DifficultyRecommended Criteria
Easy6–7
Medium8–9
Hard10–12
Demo Exec Com Chat New(4)
Difficulty is shaped by three factors: how resistant the AI character is, how complex the conversation is, and how strict the grading expectations are.
You can add scenario-specific grading notes directly in the Good/Fair/Poor text fields. For example: “User can only receive Good if they ask about timeline before discussing pricing.”

​
Troubleshooting

IssueSolution
Character reveals too muchAdd response length limits; remix to add discovery depth
Character is too difficultRemix and ask for an easier version
Conversation feels unnaturalAdd natural speech patterns; test and iterate
Missing evaluation criteriaEdit criteria directly after publishing
Info from attachment isn’t showingExplicitly tell the agent to pull in that specific information
First version didn’t feel rightPaste the session transcript into a remix and show the agent what to fix

​
Getting Help

Need help? Contact us at [email protected] for guidance on building scenarios or any questions about the Scenario Studio.