AI Labs: Learning with Situational Roleplay Training

Taylor Blake

mardi 14 avril 2026

Partager

Everything discussed in AI Labs content is experimental, in progress, and not guaranteed to become part of the Degreed product suite.

You have an important meeting in two hours that you need to prepare for. Your manager isn’t available to practice with on such short notice. You have your slide deck ready, but you aren’t sure what questions the stakeholders on the call might ask. 

This isn’t the time to wing it. A call with multiple, key stakeholders isn’t something you can just read up on and prepare for with a moment’s notice. What you are looking for is an experience that’s highly customized to your in-the-moment need, which in this case, is practice and preparation for an important conversation. This is a learning moment, but traditional learning isn’t built for it. 

This is the kind of in-the-flow learning that I wanted to see if we could use AI to provide. 

But first, let me share a little bit about AI Labs, in case you’re unfamiliar with the initiative. We experiment with the most useful ways to activate the latest AI capabilities, align them to real business needs, and see what can actually provide value in the market. A lot of what we build never makes it into the product. We test, we learn, we iterate, and we move on. 

But every once in a while, something shifts how we think about the future of learning.

Learning in the AI-Native Era

I spend a lot of time thinking about a simple question: What will learning actually look like in an AI-native world?

Traditional learning often falls short when it comes to changing behavior in the moments where performance matters most. The traditional learning model is too generic or too slow or doesn’t offer feedback to cement learning and create readiness faster. Consider the example we provided of preparing for a meeting in a couple of hours. Where are the live practice opportunities?

To meet this kind of need, I see learning becoming more:

• Specialized to the task
• Personalized to the individual
• Situational to the moment

This is the line of thinking I used to bring our AI roleplay simulations experiment to life.

AI Roleplay Conversations In Action

In this experiment, we wanted to see if we could generate specific learning experiences based on in-the-moment user needs. 

After considering what it could look like to meet situation-specific needs in practice, we created an AI agent that would allow users to type in a custom scenario (i.e., an upcoming call with investors). The agent would then generate a real simulation of that event and that would allow the learner to practice in a multi-modal format.

Let’s go back to the scenario where I have a fast-approaching meeting with stakeholders. I would go into this new tool, describe my upcoming meeting, including any information I have on who I’m presenting to and what the topic or goal is. AI would then generate a highly realistic video call interface.

Up to four AI personas can join the roleplay conversation, each with a distinct role, personality, and set of priorities that can mimic the situation and people I’m about to face. They ask questions, interrupt, and challenge you just like real stakeholders would. That real tension makes the experience even more useful.

The simulation provides user feedback both during and after the interaction, including input on how you handled questions, how clearly you communicated, and even how effective your slide deck was.

This changes how corporate training and learning is delivered. Users can run these simulations anytime. Flexibility like this matters in the real world. It’s not always realistic to practice for two weeks before an important call. Sometimes, it’s two hours before a call when you have time to squeeze in the needed preparation. 

These simulations ensure there’s always time for practice, and that high-stakes meeting never has to be your trial run.

What Makes These AI Simulations Different?

This is not just an opportunity to practice a presentation or important one-to-one interaction and get feedback. Here’s what makes this AI simulation unique:

1. Multiple personas 
   Most roleplay tools simulate a single interaction. This introduces competing perspectives, which is much closer to how real meetings unfold. These personas have distinct personalities and goals, and they ask tailored questions based on the content and their assigned perspective. 
   
2. Multi-modal input and feedback
   These simulations can also be a conversation with unexpected questions and in-depth interactions. The system evaluates all parts of the input, including what you say, how you say it, what you’re sharing live on the screen, and how it all fits in with the overall purpose and goal. 
   
3. Real-time and post-session feedback
   You get signals during the interaction so you can adjust on-the-fly. You also get more structured feedback afterward. That combination enables better in-the-flow learning, as well as reflection and iteration to improve performance.
   
4. Setup speed and complete personalization
   The experience is generated on-demand from a simple prompt. It’s not pre-built, and it’s based on your exact specifications. Need to add another persona to the call ASAP? It’s just a matter of adjusting the prompt.

The real-time feedback and opportunities to update your approach change the whole landscape of how we learn at work. 

Building the Future of Learning Through Simulations

This brings me back to the question: What does learning actually look like in an AI-native world?

These situational AI roleplay simulations can create an entirely different kind of learning system. You don’t go looking for content on a given topic. Instead, you say, “I need to practice this” or “I have this event coming up,” and the system creates the right experience for you right when you need it.

As development becomes more specialized and situational, the scope of possible learning moments expands exponentially. It’s experiential learning on demand, which is learning with a lot more potential to address the reality of constant change.

Learn more about this experiment and other ongoing innovations during my Degreed In Action webinar. Reach out to me on LinkedIn for more information on how to try these experiments for yourself.