IceGlide

• 

  Dashboard

• 

  Personalized plan

• 

  First lesson plan example

• 

  Skills list

• 

  First skill example

• 

  AI Coach chat example

• 

  AI video analysis example

Inspiration

Ice Glide was inspired by my own experience learning how to ice skate as a beginner. I realized how difficult it is to practice outside of formal lessons without clear structure or feedback. Many beginners feel lost, unsure if they are practicing correctly, and don’t have access to consistent guidance. I wanted to create a tool that makes skating more accessible, less intimidating, and more engaging. The goal was to give beginners a way to practice independently while still feeling guided and supported. Ice Glide combines structured learning, video guidance, and AI-powered feedback to recreate the experience of having a coach, but in a more flexible and accessible ($) way.

What it does

Ice Glide is a mobile-friendly web app designed to help beginner ice skaters practice in a more structured and engaging way. Users can follow guided training plans based on their level, review previous skills, and explore individual skating techniques through dedicated skill pages. Each skill includes a demonstration video, and users can also upload their own skating videos to receive feedback. Once a video is uploaded, the app analyzes the user’s movement using pose detection technology, tracking body landmarks such as joints and posture alignment. Based on this, the app provides simple, actionable feedback on posture, balance, and movement quality. In addition, the app includes an AI Coach that users can chat with for guidance, explanations, and encouragement. Ice Glide also integrates voice instructions using ElevenLabs, allowing users to listen to feedback instead of only reading it. Overall, Ice Glide transforms independent skating practice into a guided, interactive, AI-powered experience.

How we built it

Ice Glide was built using a low-code platform (Base44), focusing on rapid prototyping and iteration. The app combines multiple components into a single experience: -User authentication for personalized access -Structured training plans based on skill progression -Video upload and playback for skill practice -Pose detection (MediaPipe) for video analysis -AI Coach for conversational interaction -ElevenLabs integration for voice-based instructions The goal was to integrate AI features with a clean and intuitive user experience while keeping the application lightweight and responsive.

AI Implementation

-Pose Detection (MediaPipe): User-uploaded videos are analyzed using pose estimation to detect body landmarks such as joints and posture. This enables simple feedback on balance, posture, and movement quality. -AI Coach: A conversational assistant that provides guidance, explanations, and encouragement based on user input. -Voice AI (ElevenLabs): Text instructions are converted into natural-sounding speech using a custom voice, making the experience more interactive and accessible. To improve performance, generated audio can be cached and reused when possible.

Challenges we ran into

-Optimizing video performance and reducing load times across devices -Integrating multiple AI systems into a cohesive flow -Designing meaningful feedback without building a complex custom model -Managing time constraints while delivering a complete product

Accomplishments that we're proud of

-Finishing the app for my first Hackathon! -First time building a whole functioning product with AI -Successfully combining multiple AI technologies into one application -Building a fully functional, mobile-friendly app in a short time -Creating a beginner-friendly and accessible experience -Turning independent skating practice into a guided system

What we learned

-AI prompting -How to integrate AI tools into a real product -The importance of performance optimization for media-heavy apps -How to balance technical complexity with user experience -How to rapidly prototype and iterate under time constraints

What's next for IceGlide

-More advanced movement analysis and feedback -Progress tracking and performance insights -Expanded skill levels and content -More intelligent and personalized AI coaching features

Built With

• base44
• elevenlabs
• javascript
• mediapipe