Main Project Image

Designing Our AI Mentor

Industry

Education Tech

Client

Airlearn

Project Time

11 Weeks

About the Client

Airlearn is a popular, AI-driven mobile application designed for language learning, used by over 2 million learners globally. It provides bite-sized, interactive lessons in over 25 languages, focusing on practical conversation, cultural context, and gamified, spaced-repetition techniques to boost proficiency.

My Role

Interaction Design, Visual Design, User Flows

Key Metrics

Key Metrics

41%

41%

AI-driven features contributed to a 51% year-over-year increase in DAU.

AI-driven features contributed to a 51% year-over-year increase in DAU.

70,000

70,000

Ailrearn crossed 70,000 MAUs in late 2025, a 40% increase attributed to the scalability of AI features like automated course creation.

Ailrearn crossed 70,000 MAUs in late 2025, a 40% increase attributed to the scalability of AI features like automated course creation.

Designing an AI Mentor That Feels Human

Designing an AI Mentor That Feels Human

Designing an AI Mentor That Feels Human

The AI Mentor was conceived as more than a feature, it was designed as a learning companion embedded directly into the user’s flow. From a product design standpoint, the challenge was to make advanced AI feel approachable, non-intrusive, and emotionally supportive. Kai was intentionally positioned as a calm guide rather than an authoritative instructor, appearing contextually during moments of friction such as inactivity, hesitation, or mistakes. This ensured the mentor felt timely and helpful, not overwhelming. Every interaction was designed around real learning behaviors. The in-lesson activation logic, Magic Tab entry point, and “Ask me anything” flows were carefully orchestrated to minimize cognitive load while maximizing usefulness. The UI adapts based on intent like listening, speaking, revision, or practice so users never have to explain what they need; the system already knows. Feedback states, error handling, and even “exhausted mode” were deliberately designed to respect learner fatigue, reinforcing trust rather than pushing engagement blindly.

The AI Mentor was conceived as more than a feature, it was designed as a learning companion embedded directly into the user’s flow. From a product design standpoint, the challenge was to make advanced AI feel approachable, non-intrusive, and emotionally supportive. Kai was intentionally positioned as a calm guide rather than an authoritative instructor, appearing contextually during moments of friction such as inactivity, hesitation, or mistakes. This ensured the mentor felt timely and helpful, not overwhelming. Every interaction was designed around real learning behaviors. The in-lesson activation logic, Magic Tab entry point, and “Ask me anything” flows were carefully orchestrated to minimize cognitive load while maximizing usefulness. The UI adapts based on intent like listening, speaking, revision, or practice so users never have to explain what they need; the system already knows. Feedback states, error handling, and even “exhausted mode” were deliberately designed to respect learner fatigue, reinforcing trust rather than pushing engagement blindly.

Large Project Gallery Image #1
Large Project Gallery Image #1
Large Project Gallery Image #2
Large Project Gallery Image #2
Project Gallery Image for 50% width of the screen #1
Project Gallery Image for 50% width of the screen #1
Project Gallery Image for 50% width of the screen #2
Project Gallery Image for 50% width of the screen #2
Large Project Gallery Image #3
Large Project Gallery Image #3
Large Project Gallery Image #4
Large Project Gallery Image #4

From Prototype to Scalable System

From Prototype to Scalable System

From a development standpoint, the AI Mentor was built as a modular system that could scale across languages, lesson types, and future AI capabilities. Visual states, prompts, and mentor responses were standardized into reusable patterns, allowing rapid iteration without breaking the experience. This tight collaboration between design and engineering ensured Kai remained consistent, performant, and extensible that transforms a complex AI backend into a simple, empathetic product surface that learners could rely on daily.

Building an Award System

Good Retention with Leagues