Webinar
How to Build Realistic AI Roleplay that Actually Grows Readiness
IN THIS DEMO, YOU WILL SEE
|
1
|
Scenario creation: resolving peer conflict A custom scenario built from scratch, tailored to a meeting your team faces. |
|
2
|
Live roleplay with Anya A real-time conversation with an AI avatar who pushes back. Not a chatbot. |
|
3
|
The cohort dashboard Team-level readiness data pulled straight from learner results. |
This free AI roleplay training webinar shows L&D teams how to build a realistic conversation scenario from scratch, score it with a custom rubric, and turn every practice session into readiness data leaders can act on.
Chen Zhang, Chief Growth Officer at Rapport, built a peer conflict scenario live at ATD Demo Day. The conversation ended in agreement and felt like a win, yet the scorecard came back at 45. Watch the replay to see why.
640,000+ role plays run on Rapport to date |
6 inputs to build a full scenario |
45 score on a conversation that "went well" |
2 feedback habits that caused the gap |
KEY TAKEAWAYS
- A complete AI roleplay scenario, with persona, hidden motivations, and a grading rubric, takes a few minutes to build. No vendor ticket needed.
- Agreement is not proof of a good conversation. The live roleplay ended with a meeting booked and still scored 45.
- The two missed criteria were describing observable behavior and explaining impact. Both are habits, not knowledge gaps.
- Individual scores roll up into a cohort dashboard, so trainers can see which skill to coach next.
- Customer content is not used to train third-party AI models, and audio and full transcripts are not stored long term.
What You Will Learn in This AI Roleplay Webinar
Most soft skills training stops at knowledge. This session covers the full loop that turns practice into proof. You will see each step live, with no slides after the first few minutes.
|
1
Build Create a peer conflict scenario from scratch, including the avatar's personality and private motivations. |
2
Practice Watch a live conversation with Anya, an AI avatar who pushes back, negotiates, and reacts to tone of voice. |
3
Measure See the rubric-based scorecard, then the cohort dashboard that shows team readiness skill by skill. |
Figure 1. The three-part loop covered in the webinar: build, practice, and measure.
How to Build an AI Roleplay Scenario in Six Inputs
Scenario authoring is a task your training team owns. It does not require a services engagement. Six inputs produced the complete, reusable scenario shown in the demo.
| Input | What it sets | What the demo used |
|---|---|---|
| Scenario type | The category, which shapes the default rubric and report | Growth and development |
| Situation | What is happening and why the conversation exists | A direct report is in conflict with a peer, and delivery has slipped |
| Learner goal | What the learner is trying to achieve | Resolve the conflict as the manager |
| Avatar | Who the learner faces, including look and voice | Anya, the direct report |
| Personality traits | Tone and manner toward the learner | High standards, frustrated, quick to blame others |
| Background context | Private thoughts the avatar does not volunteer | A belief that the peer's working style is the real problem |
Figure 2. Background context is the input most teams underestimate. Real colleagues hold things back, and a persona that does the same makes practice worth doing.
Want the full walkthrough? Read how to design an AI roleplay scenario that actually changes behavior. For persona wording, see how to write a prompt for your AI avatar.
Why a Conversation That Ended Well Scored 45
Anya blamed her peer, pushed back on a three-way meeting, and agreed only once the manager kept it forward-looking. By most standards, the manager won. The rubric saw it differently.
Live roleplay scorecard: 45 overall
| Open the context | Partial credit |
| Describe the situation | Partial credit |
| Describe observable behavior | Not addressed |
| Explain the impact | Not addressed |
Figure 3. Bar length shows the result category from the learner report, not an exact sub-score.
The manager knew the situation well. She simply skipped the two statements that make feedback land:
- Observable behavior. No specific action was named, so Anya never learned which behavior was the problem.
- Impact. The cost of the missed timeline was never stated, so there was no reason to treat the change as important.
These two steps come straight from the Situation-Behavior-Impact (SBI) feedback model from the Center for Creative Leadership. We break down the pattern in why managers score lower than they think.
|
Run the same scenario yourself Face Anya, get your own scorecard, and see which criteria your conversation leaves on the table. It takes a few minutes and no setup. Try the peer conflict scenario free |
From One Score to Team Readiness: The Cohort Dashboard
Every learner's scores roll up into a dashboard. It answers three questions: how the team performs overall, how it does on one scenario, and how one learner is progressing. Comparing scenarios is the key move, because a skill that holds in one context and fails in another tells you more than any average.
| What a completion rate tells you | What readiness data tells you |
|---|---|
| Who finished the course | Which specific behavior the cohort is weakest on |
| That time was spent | Whether practice is producing improvement |
| Nothing about the next step | Which competency to coach next, and for whom |
Figure 4. Completion measures activity. Readiness measures capability.
The four readiness metrics worth tracking
- Score by criterion: which behavior the cohort is weakest on.
- Attempts per learner: whether the team is really practicing.
- First attempt to best attempt: whether practice is producing improvement.
- Cross-scenario consistency: whether the skill holds in a new context. This is the real readiness call.
Go deeper in completion rates are not readiness and how to run a skill gap analysis. Roleplays can also run inside your LMS, since Rapport is SCORM compliant.
AI Roleplay Data Security: What the Audience Asked
Data questions were the largest cluster in the live chat. Here is the short version given on the call.
✓ No model training on your content Customer conversations are not used to train third-party AI models. |
✓ No long-term audio or transcripts They are processed to build the report. Login details, user ID, and high-level scores persist. |
✓ Secure knowledge base Your content, rubrics, and scenarios drive only your scenarios and reports. |
✓ Regional hosting Servers are US-based, with EU or Asia options for enterprise customers. |
Bring the full checklist to your IT review: AI roleplay and employee data: 12 questions to ask before you roll it out.
Who Should Watch This Webinar
- L&D leaders who need to prove readiness, not just report completions.
- Instructional designers building soft skills, manager, or conflict resolution programs.
- Sales enablement teams looking to scale coaching across a distributed team.
- HR and people leaders weighing AI roleplay against manager-led practice.
More From the ATD Demo Day Series
RECAP We built an AI roleplay live at ATD Demo Day. The manager scored 45. |
FEEDBACK SKILLS Why managers score lower than they think |
SCENARIO DESIGN How to design an AI roleplay scenario that actually changes behavior |
MEASUREMENT Completion rates are not readiness |
DATA AND SECURITY AI roleplay and employee data: 12 questions to ask |
START HERE The complete guide to AI role play training |
Frequently Asked Questions About AI Roleplay Training
What is AI roleplay training?
AI roleplay training lets learners practice real conversations with an AI avatar that reacts in real time. Each session is scored against a rubric, so learners get specific feedback and leaders get objective readiness data.
How long does it take to build an AI roleplay scenario?
A few minutes. The trainer sets six inputs: scenario type, situation, learner goal, avatar, personality traits, and background context. Then they pick a grading framework and save it to a reusable library.
Can AI roleplay use our own scoring rubric?
Yes. Built-in rubrics are based on established frameworks such as SBI. Enterprise customers can upload their own rubric and score every learner against their own standard.
Does AI roleplay work inside our LMS?
Yes. Rapport is SCORM compliant, so scenarios can be embedded in your LMS. Teams can also share a private link or use the learner portal with single sign-on.
Is learner data used to train AI models?
No. Customer conversations are not used to train third-party AI models. Audio and full transcripts are not stored long term.
Do managers see every learner's roleplay report?
Not by default. Individual reports stay with the learner to protect psychological safety. Aggregate scores roll up to the cohort dashboard.
Is this AI roleplay webinar free to watch?
Yes. The full recording is free on this page. You can also try the same peer conflict scenario from the demo for free.
|
Find out what your managers would score. Run the scenario from the webinar, or see how Rapport fits your L&D program. Try the scenario free Book a demo |