Our MentorAI Story

What Inspired Us

Throughout my academic journey, I have often felt that conventional classroom learning can be deeply unjust for many students. The traditional model—one teacher, many students, and a single, uniform pace—rarely aligns with the varied learning curves and needs of individual learners. I have seen firsthand how some students are left behind while others are held back, simply because the system is designed for the "average" rather than the individual. This realization inspired me to find a way to personalize education and truly empower every learner, regardless of their background or starting point.

What We Learned

Building MentorAI was a profound learning experience. We delved into the capabilities of large language models (LLMs) and saw how they could be orchestrated to act as adaptive teaching agents. Through researching best practices in personalized education, we learned how crucial it is for students to feel seen and supported according to their unique abilities, languages, and learning speeds. We also explored the immense power of voice and language adaptation in breaking down barriers for learners in rural and under-resourced communities.

How We Built the Project

We set out to build an AI-powered teaching platform that assesses each student individually, then adapts the curriculum in real time to suit their needs:

• Assessment: The learning journey starts with an entrance exam, delivered in the student's preferred language and through voice, which gauges their current proficiency and identifies learning gaps.
• Personalized Curriculum: Based on the assessment, MentorAI’s intelligent agents construct a custom lesson plan tailored to the student’s strengths, weaknesses, and pace.
• Adaptive Delivery: Lessons are delivered both as text and voice, further enhancing accessibility for diverse learners.
• Interactive Quizzing & Feedback: Quizzes adapt to student progress and provide immediate, personalized feedback. This ensures students are always working at a level that’s right for them.
• Continuous Adaptation: The platform continuously updates the learning path according to quiz outcomes and feedback, ensuring mastery before moving forward.
• Teacher Dashboard: Educators can view detailed progress charts and download reports, allowing them to support students more effectively.

As a demonstration, we focused on the English language subject, showcasing how MentorAI adapts its course structure and teaching style for each student’s needs.

Challenges We Faced

Personalizing education with AI is an ambitious challenge. Some of the obstacles we encountered included:

• Balancing Adaptivity and Structure: Ensuring that the AI adapts to individual needs without losing sight of curriculum standards.
• Multilingual Support: Integrating robust voice and text support across diverse languages proved technically complex.
• Feedback Loop Design: Creating meaningful, constructive feedback that encourages students without overwhelming them requires careful design.
• Scalability: Making sure the platform can be used by many students and teachers simultaneously, while maintaining speed and reliability.

Despite these challenges, our passion for educational equity kept us going. We believe MentorAI is a step toward more just, personalized, and empowering learning for all.

Built With

• and-whisper-(voice/text-to-speech-and-speech-to-text)-firebase-auth-and-flask-(authentication)-vappi-api-mocks-(mock-backend-and-lms-integration)-plotly-and-matplotlib-(data-visualization)-deployed-for-local-use
• anthropic
• fetch.ai
• google-tts
• groq
• orkes-conductor-(agent-orchestration-and-workflow)-letta-pro
• python
• react
• streamlit
• supabase
• vappi-api