A Product Design Case Study

Picture this: You’ve spent six months using a language app, completed hundreds of lessons, and mastered verb conjugations. Then you land in Paris, walk into a café, and freeze. The barista asks a simple question. Your brain scrambles. The words you practiced vanish. You point at the menu and mumble “café, s’il vous plaît.”

Sound familiar? You’re not alone. This gap between classroom fluency and real-world conversation has plagued language learners for decades. But in the past two years, a new breed of AI-powered tools is finally bridging that chasm — and the secret ingredient isn’t better grammar engines or flashier gamification. It’s context.

The Scenario Revolution

In 2024, Duolingo introduced Video Call with Lily, its AI character who adapts conversations based on your proficiency level, guiding you through ordering coffee, asking for directions, or discussing vacation plans. By September, the feature had helped drive a 50% increase in subscriptions. But Duolingo isn’t alone. Mondly’s VR chatbot drops you into immersive scenarios — checking into a hotel, navigating an airport. Babbel Speak launched “Everyday Conversations” that simulate real-life dialogues with AI-powered prompts. Even traditional platforms are pivoting: context is king.

What’s changed? Large language models like GPT-4 have evolved beyond pattern recognition into genuine conversational intelligence. These chatbots don’t just respond — they adapt, remember context, and adjust complexity on the fly. More importantly, they’re finally doing what language teachers have known for years: the best way to learn a language is to use it in situations that mirror real life.

Why Scenarios Work (And Flashcards Don’t)

The research is compelling. A 2024 meta-analysis of 70 effect sizes from 28 studies found that chatbot-assisted language learning produced a significant positive effect (g = 0.484) compared to traditional methods. But here’s the kicker: not all chatbot interactions are created equal.

Traditional language apps focus on decontextualized drill-and-practice. You learn “Où est la gare?” but have no mental model of when or why you’d say it. Scenario-based chatbots flip this model. Instead of aimless banter, scenarios guide conversations toward specific objectives, like declining a request or ordering food. The learning objective is embedded in a situation you’ll actually encounter.

Think of it as the difference between knowing the word “reservation” and having practiced the entire negotiation: “I’d like to change my reservation. Is there availability for tomorrow instead? Great, and can I request a room with a view?”

Digital scenario-based teaching creates immersive learning experiences using multimodal resources like videos, audio, and interactive tasks, which spark learners’ interest and enhance linguistic proficiency. When you practice language in context, your brain doesn’t just store vocabulary — it builds neural pathways connecting words to situations, emotions, and outcomes.

The Hidden Advantage: Anxiety Reduction

Here’s where things get interesting from a psychological standpoint. AI-powered chatbots create safe, judgment-free environments where learners can engage without the anxiety that often accompanies speaking with native speakers. A recent study on AI conversation bots found they significantly reduced foreign language speaking anxiety while simultaneously improving speaking skills.

Why does this matter? Language anxiety is one of the biggest barriers to fluency. We’ve all been there — the heart-pounding moment before speaking in a foreign tongue, the fear of making mistakes, the embarrassment of mispronunciation. When chatbots offer supportive and informative feedback rather than harsh criticism, learners experience positive emotions like motivation and a sense of progress.

Duolingo’s Lily doesn’t judge when you stumble over subjunctive tenses. Babbel Speak doesn’t sigh when you ask it to repeat something for the fifth time. This low-pressure practice environment builds what psychologists call “self-efficacy” — the belief that you can succeed in real conversations.

The Design Principles That Separate Winners from Losers

Not all scenario-based chatbots are effective. After analyzing dozens of implementations, several design principles emerge as critical:

Adaptive Complexity: Learning objectives must align with proficiency levels, with scenarios calibrated to where learners are in their course progression. A beginner needs different scaffolding than an intermediate learner attempting business negotiations.

Authentic Contexts: The best implementations don’t invent arbitrary scenarios. They mirror real-world situations learners will actually encounter: travel, work meetings, social interactions. Babbel’s “Everyday Conversations” feature covers realistic situations like introducing yourself, talking to friends after work, or ordering food.

Immediate, Constructive Feedback: After interactions, learners receive AI-powered feedback on accuracy and complexity, plus tips for future conversations. This isn’t about red-pen corrections — it’s about metacognitive awareness of how you communicate.

Multimodal Engagement: Voice recognition matters. Studies on voice-based AI chatbots show they help learners practice pronunciation and develop confidence in speaking. Text-only interfaces miss a critical dimension of language learning.

The Product Design Challenge

For product designers and educators, scenario-based chatbots present fascinating UX challenges. Duolingo injects every prompt with a scenario that sets the scene — the setting, the character’s role, what they want to do, and an appropriate learning objective. This requires enormous content creation, quality control, and pedagogical expertise.

The engineering complexity is equally daunting. Building video calling AI proved more involved than other features, requiring careful tuning to maintain conversational quality while preventing the bot from becoming too predictable or scripted. And then there’s the language quality problem: English chatbots perform better than those in less-resourced languages, creating equity concerns.

But here’s the strategic opportunity: Duolingo’s move to become “AI-first” has helped drive 50% subscription growth, with users specifically citing the Lily chatbot as a key value driver. The companies that nail scenario-based learning aren’t just building better education products — they’re creating moats that generic LLM competitors can’t easily replicate.

What This Means for Designers and Educators

The shift toward scenario-based AI chatbots signals three major implications:

1. Context is the New Content
The winners won’t be those with the most lessons or the biggest vocabulary databases. They’ll be the ones who design the most authentic, psychologically safe scenarios for practice. Product teams need linguists, instructional designers, and cognitive scientists at the table — not just engineers.

2. Personalization Goes Deeper Than Difficulty
LLM-based chatbots significantly improve both receptive and productive vocabulary knowledge and contribute to long-term retention. But the real magic happens when systems remember your goals, track your anxiety triggers, and adapt scenarios to your learning style. We’re moving from “adaptive difficulty” to “adaptive context.”

3. The Human Touch Remains Irreplaceable
Humans write the scenarios, ensure initial prompts align with course progression, craft initial messages, and constantly review AI-generated content for factual accuracy and appropriate tone. The most successful implementations combine AI’s scalability with human expertise in pedagogy and cultural nuance.

The Road Ahead

Scenario-based chatbots aren’t perfect. They still face challenges like restricted emotional awareness, deficiency in genuine human interaction, and potential bias reinforcement. Voice recognition quality varies across languages. Privacy concerns remain, especially for younger learners. And no AI can replace the cultural richness of conversation with a native speaker.

But for the first time, millions of learners have access to something that previously required expensive tutors or immersion programs: unlimited, judgment-free practice in realistic situations.

The question isn’t whether AI chatbots will transform language learning — they already have. The question is whether product designers and educators will rise to the challenge of designing experiences that honor both the power of technology and the deeply human nature of communication.

Because ultimately, language learning isn’t about memorizing words. It’s about building the confidence to walk into that café in Paris, look the barista in the eye, and have a real conversation. Scenario-based AI chatbots are finally giving millions of people the practice they need to make that happen.

References:

1. Duolingo. (2023). “Introducing Duolingo Max, a learning experience powered by GPT-4.” Duolingo Blog. https://blog.duolingo.com/duolingo-max/
2. Wang, F., Cheung, A. C. K., Neitzel, A. J., & Chai, C. S. (2024). “Does chatting with chatbots improve language learning performance? A meta-analysis of chatbot-assisted language learning.” Review of Educational Research. https://doi.org/10.1102/00346543241255621
3. Design language learning with artificial intelligence (AI) chatbots based on activity theory. (2025). Smart Learning Environments. https://slejournal.springeropen.com/articles/10.1186/s40561-025-00379-0
4. Investigating the role of AI-powered conversation bots in enhancing L2 speaking skills. (2025). Humanities and Social Sciences Communications. https://www.nature.com/articles/s41599-025-05550-z
5. The use of artificially intelligent chatbots in English language learning: A systematic meta-synthesis study. (2024). ReCALL, Cambridge Core.
6. The promises and challenges of AI-based chatbots in language education through the lens of learner emotions. (2024). PMC. https://pmc.ncbi.nlm.nih.gov/articles/PMC11416278/
7. The impact of chatbots based on large language models on second language vocabulary acquisition. (2024). Heliyon. https://pmc.ncbi.nlm.nih.gov/articles/PMC10850600/
8. Evaluation of the Implementation Effectiveness of Digital Scenario-based Teaching in University-level English Conversation Instruction. (2025). Research Square. https://www.researchsquare.com/article/rs-6437895/v1
9. Babbel. “Learn with your own voice: Babbel launches two new speech-based features.” https://www.babbel.com/press/
10. Duolingo. (2025). “Can I use ChatGPT to practice a new language?” Duolingo Blog. https://blog.duolingo.com/chatbot-language-practice/
11. The 74 Million. (2025). “As Duolingo Turns to AI, Some Users Say Language App Has Joined ‘The Dark Side’.” https://www.the74million.org/

Scenario-based AI Chatbots for Language Learning was originally published in UX Planet on Medium, where people are continuing the conversation by highlighting and responding to this story.