I designed this simulation knowing that I didn’t want to simply drop the learner into a conversation with a chatbot, especially around such a sensitive topic and challenging interpersonal skill. Instead, I used narrative, context, and emotional setup to make the experience feel like a believable real-world interaction rather than just a chatbot exercise.

This design decision was intended to help learners understand why the conversation matters and to establish the emotional stakes of the interaction before the simulation begins. By grounding the experience in a realistic workplace scenario, I aimed to create a stronger sense of immersion and encourage more thoughtful, authentic learner responses.

Key design decisions included:

• Creating a realistic office setting

• Using cinematic intro videos

• Including workplace chatter and social context

• Providing emotional context around Mira’s situation

• Designing a realistic texting interface

Together, these elements created a clear reason for the learner to enter the conversation and helped support a more immersive, emotionally authentic learning experience.

I wanted the conversation with Mira to feel dynamic and responsive rather than scripted. Because emotional intelligence depends heavily on context, tone, and interpersonal judgment, I designed the experience so that the conversation changes based on the learner’s responses and communication style. This allows each learner to practice their emotional intelligence skills and experience different conversational outcomes depending on the decisions they make throughout the interaction.

A traditional branching scenario would have significantly limited learner choice and required pre-written response paths that could unintentionally guide learners toward the “correct” answer too easily. By using AI-powered conversation instead, I was able to create a more flexible and authentic interaction that encourages learners to think critically and respond naturally in the moment.

The learner feedback and scoring system are also AI-powered. Rather than receiving generic right-or-wrong feedback, learners receive tailored feedback based on the specific communication choices they made during the conversation. This allows the experience to feel more personalized, reflective, and meaningful across multiple attempts.

AI Tools Used:

• Devlin.ai was used to create the AI-powered conversation and feedback experience with Mira.

Using AI in this way allowed me to create a richer and more realistic practice experience where learners can test their communication skills, reflect on the outcomes of their choices, and refine their approach through repeated practice.

To make the experience feel more emotionally engaging and dynamic, I designed multiple emotional states for the Mira character that change throughout the conversation based on learner interactions. As the learner communicates with Mira, her facial expressions, body language, and overall emotional presentation shift to reflect the emotional progression of the interaction.

This design decision was intended to reinforce the interpersonal impact of the learner’s communication choices. Because emotional intelligence relies heavily on recognizing emotional cues, the visual changes help learners interpret how their responses are affecting Mira emotionally rather than relying on text alone.

How I accomplished this:

• Used Storyline object states, variables, and triggers to manage Mira’s emotional state changes

• Used Devlin.ai to change the Storyline variables that triggered changes in Mira's state.

• Used AI-assisted visual asset creation to design multiple emotional versions of the Mira character

This approach increased immersion, strengthened the storytelling, and helped create a more emotionally authentic simulation experience.

Creating the simulation experience I envisioned required extensive prompt engineering, testing, and refinement. Rather than treating AI as a standalone solution, I approached the prompting process as an instructional design challenge—continually aligning the AI's behavior with the learning objectives, audience needs, and overall goals of the project.

Throughout development, I carefully defined how Mira should communicate, how quickly trust should develop, what behaviors should be rewarded, and how challenging the interaction should feel for learners. These decisions helped ensure that the conversation remained believable, emotionally authentic, and instructionally meaningful.

User testing played a critical role in this process. Early feedback revealed that Mira was often too guarded, making it difficult for learners to build trust and progress the conversation. Through multiple rounds of iteration, I refined the prompts to create a more balanced experience that remained realistic while providing learners with meaningful opportunities to practice emotionally intelligent communication.

This iterative approach helped shape an AI experience that feels human-centered, supports the learning objectives, and encourages learners to reflect on the impact of their communication choices.

To balance challenge with support, I intentionally incorporated learner scaffolding throughout the experience. Because emotionally intelligent communication can feel nuanced and difficult to navigate, I included an optional help button that learners can access at any point during the simulation. Rather than providing answers, the help guide offers practical communication strategies, including what to do, what to avoid, and helpful reminders for building trust and supporting Mira effectively.

At the same time, learner autonomy was an important consideration. The help guide is entirely optional, allowing learners to decide how much support they want or need during each attempt. This creates opportunities for both independent exploration and guided practice.

I also introduced intentional constraints to create a more realistic practice environment. Learners are limited to 10 responses, reflecting the reality that difficult workplace conversations often require thoughtful communication within a limited amount of time. This constraint encourages learners to be intentional with their responses while keeping the experience focused and manageable.

Support for the learner was also built directly into Mira's character design and AI prompting. Mira was intentionally designed as a high-performing employee who struggles with vulnerability, asking for help, and feeling like a burden to others. While emotionally guarded, she was also designed to provide subtle cues and hints about what she may be feeling or needing throughout the conversation. These cues create natural opportunities for learners to demonstrate empathy, ask thoughtful follow-up questions, and practice recognizing emotional signals that might otherwise go unspoken in real workplace conversations.

Together, these scaffolds and constraints helped create an experience that is challenging enough to promote skill development while providing learners with the support needed to remain engaged and successful.