PosePal

Inspiration

A friend of ours avoided the gym because of body-shaming experiences. He started training at home but lacked confidence, guidance, and accountability. He told us: “Apps just count reps, but they don’t actually see me. They don’t correct me, they don’t push me.” That frustration inspired Posepal: a coach that not only tracks your movements in real-time but also speaks to you with empathy, energy, and personality, removing the loneliness of solo workouts.

What it does

Posepal is an AI-powered fitness coach that uses computer vision and real-time feedback to guide workouts directly in your browser. It tracks your form, counts reps, detects mistakes, and provides instant, persona-based voice coaching. Users can choose between different motivational styles (Default, Goggins, Barbie) and see progress in a workout dashboard. The result: safe, personalized, and motivating workouts without needing an expensive trainer.

How we built it

We started by defining the problem we wanted to solve: making at-home workouts safer and more motivating, especially for people without access to a coach. From there, our process looked like this: Research & Brainstorming Prototype Pose Tracking Feedback System Voice Personas LLM Integration Frontend & Dashboard User Authentication & Storage

Challenges we ran into

Latency: Ensuring real-time feedback while handling both CV and LLM processing. Voice Integration: Mapping personas to TTS voices in a way that felt natural. Data Pipeline: Sending exercise data to the LLM while keeping responses concise and relevant. Frontend Simplicity: Making it run in-browser without heavy dependencies. Motivation Design: Finding the right balance between fun personas and serious form corrections.

Accomplishments that we're proud of

Built a working prototype with live tracking, voice feedback, and a dashboard. Achieved real-time performance with Mediapipe + FastAPI + local TTS. Designed a system where AI feedback feels personal and dynamic, not generic. Create a foundation that can scale into training plans, gamification, and rehab support.

What we learned

How to balance computer vision accuracy with user experience speed. The importance of tone and persona in keeping users motivated. That accessibility and inclusivity are as important as raw performance in fitness tech. How to orchestrate multiple AI components (CV + LLM + TTS) into a seamless experience.

What's next for PosePal

Built With

• css3
• fastapi
• html5
• javascript
• llm
• python
• tts