Designing an AI-Powered Math Learning Experience

27 Nov, 2025

When the global pandemic forced an unprecedented shift to remote learning, the fragility of our educational infrastructure became starkly apparent. Our mission was born from this crisis: to transform a powerful but nascent AI technology into a personalized, scalable learning ecosystem that could bridge the widening gaps in mathematics education for students left without one-on-one support.

This initiative resulted in an integrated solution: PrepAnywhere, the core AI math tutor, and PrepBox, an interactive iPad application designed for hands-on problem-solving. Together, they form a comprehensive ecosystem engineered to support students, empower teachers, and engage parents through the complexities of modern mathematics education.

As the UI/UX Designer, I led the design process from initial user research and MVP development to final market launch. My role involved orchestrating a cross-functional collaboration with a dedicated team of engineers, data scientists, and educators from institutions like the University of Waterloo and the University of Toronto, guiding the transformation of a raw AI model into a user-centered, market-ready product.

The project's success is demonstrated by its significant market impact and business growth. The user-centric design strategy was a key driver behind these core outcomes:

A 65% increase in company revenue.
Acquisition of over 600,000 new users and an eightfold increase in AI plan signups.
Market expansion reaching 58.25 million users following a U.S. launch.
Successful fundraising of approximately five million dollars.

This case study provides a deeper analysis of the core educational challenge that necessitated this solution and the user-centered design process employed to solve it.

The Challenge: Reimagining Remote Math Education

To build a truly impactful product, it was strategically vital to understand the market context. The project was initiated in direct response to the global shift to remote learning during the COVID-19 pandemic, a period that exposed profound weaknesses in traditional educational models. The lack of in-person instruction created significant gaps in personalized student support, particularly in a subject as cumulative and conceptually demanding as mathematics.

The core problem was the widespread lack of individualized attention. Without a teacher's immediate support, students with varying learning gaps struggled to build foundational confidence. At the project's outset, our AI model was like "a young student with limited understanding"; it had to be meticulously taught before it could teach others. Our challenge was to mature this technology into a robust tool that could replicate the benefits of one-on-one tutoring in a scalable, accessible digital format.

To address this effectively, the design was grounded in a clear understanding of our target audience:

Students: Ranging from K-12 to university level, requiring on-demand homework assistance, deep conceptual understanding, and effective exam preparation.
Teachers: Seeking tools to generate worksheets aligned with US, Canadian, and Singapore Math curricula, supplement classroom instruction, and free up time from answering routine questions.
Parents: Looking for reliable resources to support their children's mathematical development outside of school hours.

With these user needs in mind, the project goal was clearly defined: to transform the company's powerful AI model into a marketable, user-centered product that provided accessible, personalized, and effective math tutoring for learners at every level. This definition of the problem laid the groundwork for a methodical design process focused on delivering a truly impactful solution.

A User-Centered Design Strategy

To create an EdTech product that was not just technologically advanced but also pedagogically sound, a user-centered design methodology was critical. The entire process was grounded in an empathetic understanding of learner behaviors, educator needs, and business goals, ensuring every decision was informed by research, testing, and stakeholder alignment.

The multi-stage design and development process followed a proven framework for building successful digital products:

Stakeholder Collaboration & Research: The project began with intensive collaboration, partnering closely with the CTO, engineers, data scientists, and educators from institutions like the University of Waterloo and the University of Toronto. We then conducted user interviews with students and teachers to identify key pain points and validate assumptions, ensuring the final product would solve real-world problems.
Information Architecture (IA) & User Flow: We translated user insights into a logical structure by creating detailed wireframes and user flow diagrams. This foundational work was crucial for optimizing the learning management system (LMS), ensuring users could navigate intuitively and find the support they needed without friction.
Prototyping & MVP Development: To test concepts and gather feedback rapidly, we developed a series of Minimum Viable Product (MVP) prototypes. These interactive models were essential for integrating and validating new features. The prototypes were subjected to rigorous usability testing, allowing us to observe user interactions firsthand and gather actionable feedback for refinement.
Iterative Design & Brand Consistency: The design process was iterative, not linear. Based on feedback from usability testing and insights from analytics, we continuously refined the product's UI and UX. Concurrently, we created and maintained comprehensive design guidelines to ensure brand consistency across all platforms, ensuring the product evolved in lockstep with user expectations.

This structured, user-focused approach enabled the team to translate a powerful AI technology into an intuitive and highly effective final product.

An Integrated and Personalized Learning Ecosystem

The final solution is an integrated ecosystem uniting the PrepAnywhere AI tutor with the interactive PrepBox application. Its success lies not in a single feature but in a strategic blend of AI technology, hands-on tools, and a multi-layered support system. Guided by core Learning Experience Design (LXD) principles, this ecosystem was engineered to be personalized, engaging, and highly effective for a diverse range of learners.

Core Platform Features & LXD Principles

AI-Powered Tutoring (“Goo”): An AI model that analyzes textbook questions (from photos) and provides similar example problems, step-by-step solutions, and conceptual video recommendations. Personalization & Scaffolding — Named “Goo,” the AI provides tailored support based on the user’s specific query, acting as a personal math coach. It doesn’t just give the answer but guides understanding, a key scaffolding technique.
Interactive Digital Whiteboard: An iPad-only app feature where students solve problems by showing their work. Tutors can view the whiteboard live to check progress and offer real-time assistance. Active Learning — This feature promotes the company’s motto of “learning by doing,” moving students from passive content consumption to active problem-solving, which is crucial for building deep conceptual mastery.
Multi-Layered Support System: A combination of a vast library of lecture and solution videos, AI-generated hints, and on-demand access to live human tutors via a “raise-hand” button. Scaffolding & Flexibility — This creates a comprehensive support network that caters to diverse learning needs, allowing students to choose the level of assistance they require, from self-service videos to direct human interaction.
Automated Assessment & Instant Feedback: Assignments submitted on the platform are graded instantly by AI, with a color-coded system (red for wrong, yellow for missed steps/incomplete, and green for correct) indicating performance. Feedback & Assessment — This provides an immediate feedback loop critical for effective learning, allowing students to quickly identify and correct misunderstandings without waiting for a teacher’s review.
Teacher & Educator Tools: Functionality allowing educators to generate custom worksheets with integrated QR codes that link directly to solution and learning video playlists. Accessibility & Curriculum Alignment — This extends the platform’s value beyond the student, making it a powerful classroom tool for educators that aligns directly with US, Canadian, and Singapore Math curricula.

The thoughtful implementation of these features resulted in a powerful platform that generated significant and measurable business and user impact.

The Impact

In product design, success must be validated by data. The value of the user-centered approach applied to the PrepAnywhere and PrepBox platforms was clearly demonstrated through significant, measurable growth in key business metrics, confirming that our design enhancements were directly correlated with the platform’s rapid ascent in the competitive EdTech market.

The project’s achievements underscore the power of aligning user experience with business strategy:

Significant Revenue and User Growth: The platform redesign and feature enhancements directly contributed to a 65% increase in company revenue and successfully attracted over 600,000 new users, demonstrating strong market adoption and commercial viability.
Increased Product Adoption: We saw an eightfold increase in AI plan signups, a critical metric confirming that the AI-powered tutoring features — the core of our innovation — were not only well-designed but also highly valued by our target audience, proving strong product-market fit.
Successful Market Penetration: The platform’s strategic design facilitated a successful launch in the United States, expanding its reach to an impactful 58.25 million users and establishing it as a key player in the North American market.
Secured Investor Confidence: A polished user experience and compelling product narrative were instrumental in stakeholder communications and marketing efforts, which successfully raised approximately five million dollars in funding to fuel further growth.

Metrics, revenue, users, funding info by CEO Min Lee

A successful product strategy where thoughtful, user-centered design served as the critical engine for business growth and market leadership.

Key Learnings and Future Implications

The PrepAnywhere project successfully transformed a powerful AI technology into a market-leading EdTech product by maintaining a relentless focus on the user experience. By deeply understanding the needs of students, teachers, and parents, we designed an integrated learning ecosystem that provided personalized, effective, and accessible mathematics support. This case study demonstrates that a user-centered design philosophy is the foundational pillar for creating products that achieve both educational impact and commercial success.

The journey yielded several critical insights applicable to the broader field of educational technology design:

The Power of Blended Intelligence: The most effective learning environment combines the scalability of AI for instant feedback and personalization with the nuance and empathy of human tutors for complex problem-solving and encouragement.
Active Learning Drives Engagement: Features that promote active participation, such as the interactive digital whiteboard, are demonstrably more effective at building deep conceptual understanding. Our commitment to a “learning by doing” philosophy was essential for fostering genuine mastery.
An Ecosystem Approach is Essential: A truly valuable educational tool cannot exist in a vacuum. By designing a platform that serves the interconnected needs of students, teachers, and parents, we created a more resilient and integrated solution that embeds itself seamlessly into the broader learning journey.
Community Fosters Persistence: Technology can feel isolating. By integrating community-oriented features like a “Discord server for a classroom-type feeling,” we learned that fostering peer-to-peer interaction and a sense of belonging is a powerful motivator that complements AI and human-led instruction.

Page updated

Google Sites

Report abuse