NOW

No gym? No problem. Static calisthenics builds insane strength and mobility at home and "NOW" gamifies your training so every hold feels like unlocking a new superpower! So? What's your excuse?

Comment

Inspiration (How NOW came to be)

In my second year of university, I finally decided to start hitting the gym. But let’s be real, I was intimidated. People younger than me were casually benching what I couldn’t even squat. That’s when I found calisthenics.

No weights. No fancy machines. Just your body vs. gravity. It builds insane strength, but it’s also risky if your form is off. And the internet is full of half-baked tutorials. I wanted a way to learn safely, track my progress, and keep myself motivated but I had no other option but to follow those tutorials. Yes, I did eventually learn but I ended up wasting a lot of time on learning.

That’s why, today, I decided to build NOW. Think of it as your AI-powered training buddy that teaches you calisthenics step by step, checks your form, and cheers you on when you unlock a new skill.

What it does (NOW, for now)

NOW makes calisthenics feel like a game where your body is the controller. Here’s how:

📚 Roadmaps for progress

  • Push skills: Elbow Lever → L-Sit → Planche (with sub-skill progressions).
  • Pull skills: Back Lever → Front Lever (with sub-skill progressions).
  • Each skill has a guide explaining the movement, strength focus, and form tips.

🗓️ Weekly training plans

  • Step-by-step goals so you’re never left wondering “what do I train today?”

🤖 AI-powered form checks

  • Upload a photo of your hold.
  • Pose detection maps your joints.
  • Angles are calculated with vector math.
  • Gemini gives personalized feedback (like a coach who actually knows biomechanics).

🏆 Gamified progress tracking

  • Earn scores for each skill (pass with ≥65%).
  • Log training notes and hold durations.
  • Collect achievements and streaks.
  • Celebrate every unlock with confetti 🎉.

How we built it

  • Frontend: Next.js 15, Tailwind CSS
  • Backend: FastAPI (Python).
  • Computer Vision: MediaPipe + OpenCV for joint detection.
  • Math Engine: NumPy.
  • RAG: Gemini 2.5 Flash.

Challenges we ran into

I initially wanted to fine-tune a lightweight open-source model for on-device analysis (so it could even run on a phone). But as a solo hacker, I had to pivot. Instead, I integrated Gemini with a structured RAG knowledge base. This let me deliver reliable coaching while still hitting my hackathon deadline.

Cutting features was painful (bye-bye real-time video analysis 😢), but it made the final product way more solid.

Accomplishments that we're proud of

  • As a one-person team, I managed to ship a working full-stack prototype in a weekend. I also learned how to set realistic milestones and prioritize features under pressure. Honestly, I’m just proud that what started as a personal problem turned into something real that others can potentially use.

What we learned

Not just technical... but life lessons:

  • 💤 The importance of sleep and energy management during hackathons.
  • 🥗 How much diet affects focus and productivity.
  • 🎯 The value of scoping down to a functional, polished MVP instead of chasing every cool feature.

What's next for Now

This started as a tool just for myself, but it’s grown into something bigger. I would want to add authentication and a real-time database, expand this into a mobile app, enable real-time video analysis, and publish online so anyone can use it, track progress, and share their journey.

🏋️ Long-term: build a community-driven calisthenics platform with challenges, leaderboards, and social features.

Built With

Share this project:

Updates