📌 Interested in Roleplay Sessions? Head to the Roleplay Sessions page to learn more and get in touch with our team.
A Roleplay Session lets learners practice a real conversation with an AI avatar and get automated feedback on how they did. You build it from three scene types — scripted, roleplay, and feedback — and then publish it to the learner portal. This article covers the full process from start to finish.
Before you start
To create a Roleplay Session, you need:
Enterprise plan access
Organization admin or creator access
A clear understanding of the skill you want to teach
Rubrics or skill criteria you want to assess learners on
If you are new to roleplay videos, start by reviewing the scene types first. This helps you understand how scripted, roleplay, and feedback scenes work together.
How to create a Roleplay Session
There are three ways to create a Roleplay Session. Choose the one that fits how much you want to build from scratch:
Duplicate a ready-made template — the fastest way. You see a complete roleplay in action and edit it to fit your scenario.
Create with Assistant — describe your scenario in a prompt, and Assistant builds the whole roleplay for you.
Add a Roleplay scene — add a Roleplay scene to a new or existing video and configure it in the editor.
Duplicate a ready-made template
The fastest way to create a Roleplay Session is to duplicate a ready-made template. A template gives you a complete roleplay in action — scripted, roleplay, and feedback scenes already built — so you can see how the pieces fit together before you change anything.
Choose a template that matches your scenario, such as a sales template for a sales roleplay.
Duplicate the template.
Edit the persona, scenario, objections, skills, and scripts to fit your needs.
Create a roleplay with Assistant
Assistant builds a complete Roleplay Session from a prompt. You describe the scenario, and Assistant creates a ready-to-try roleplay for you.
From the home page, select Create a roleplay. You can also start from the Create a roleplay entry point in the video assistant view.
Choose one of the five activities: Cold call, Discovery call, Field sales, Customer conversation, or Manager feedback.
Enter a prompt that describes the scenario, audience, and topic.
Assistant builds a full storyboard — scripted, roleplay, and feedback scenes — from the stock template mapped to that activity. It also creates a custom persona (avatar, voice, personality, and background), scenario instructions, and objections, and assigns skills from Synthesia's stock skill library.
Try the roleplay by previewing. You can manually edit any generated element, such as the persona, scenario, or skills.
✍️ What to know about Assistant-led creation.
Each activity maps to one fixed stock template. You cannot choose a different template for an activity.
Assistant selects from Synthesia's stock templates and skills. To use a custom template or a custom skill, add a Roleplay scene and build it in the editor instead.
Assistant is available only during initial creation. Once you are in the storyboard or editor, all further edits are manual.
Add a Roleplay scene to a new or existing video
For full control over each scene, add a Roleplay scene to a new or existing video and build it in the editor.
1. Design the roleplay scenario
Before you configure the scene, define what learners will practice and how they will be assessed. This helps your teaching content map directly to the roleplay.
Decide:
AI agent persona — the person your agent roleplays as — their name, role, personality, and background.
Instructions — the persona’s context and how they should respond to different questions, approaches, and signals from the learner.
Objections — the realistic pushback the avatar should raise during the conversation.
Skills — the 2–5 skills or criteria the learner should be scored on.
Time limit — the maximum duration the roleplay can run. Most roleplays are 5–10 minutes.
For detail on each field, see How do I configure my roleplay and feedback scenes?
2. Add and configure the roleplay scene
In a new or existing video, add a Roleplay scene, then fill in the configuration fields for the scenario you designed.
Scene-type menu
The roleplay scene is the core of the experience. This is where learners have a real-time conversation with the AI avatar. The avatar follows your scenario, responds to what the learner says, and raises the objections or challenges you define.
Roleplay scene configuration panel in the Synthesia editor.
3. Add the feedback scene
Add a feedback scene after the roleplay scene. Synthesia generates most of the feedback scene for you, and you set how it works.
Set the passing score, then choose whether to turn Retries on or off. If retries are on, learners can retake the roleplay and try to improve their score. If retries are off, learners get one attempt. The feedback scene appears after the roleplay and shows learners what they did well, where they can improve, and how they performed against the skill criteria.
Feedback scene settings
4. Add scripted scenes
Scripted scenes teach the skill and show what good looks like before learners practice. They often include examples, best practices, or a model answer.
Place scripted scenes before the roleplay scene. Write them after you configure the roleplay so the script maps to what learners will practice and what the feedback scene will assess.
Keep scripted content short so learners get to the practice quickly. Aim for 2–5 minutes of scripted content in total.
Scripted scene
📚 Want to go further? Learn how to configure your roleplay and feedback scenes.
Preview, generate, publish, and assign
These steps apply to all three creation methods.
1. Preview the experience
Use draft preview to run through the roleplay as a learner. Previewing helps you catch configuration issues before you publish.
During preview, check that:
The scripted scenes are clear and relevant.
The persona behaves as intended.
The objections feel realistic.
The roleplay flow makes sense.
The feedback scene scores the right skills.
The script maps to what the roleplay tests.
Fix any issues you find before you generate or publish.
Draft preview
2. Generate the video
When your configuration is complete, select Generate.
Synthesia checks for configuration errors, such as missing skill rubrics or unlinked scenes, before generation starts. If errors appear, fix them and try again.
Generation takes a few minutes.
3. Publish to the learner portal
When the video is ready, publish it so learners can access it.
Publishing requires organization admin access. After you publish, the roleplay video appears in the learner portal and is ready to share or assign.
Publish to the learner portal.
4. Share or assign it to learners
After you publish, share the roleplay video with learners. Depending on your setup, you can share a link, invite learners by email, or assign the roleplay to specific learners or teams.
Learners move through the full experience — scripted, then roleplay, then feedback — in one flow. Use learner progress or analytics to track completion, pass rates, and improvement over time.