Our Story

Inspiration

We were all athletes throughout high school — rowing, track, and golf. And even though we had great coaches, they couldn’t always be there. Most training happened alone, without real-time feedback. On big teams, coaches simply didn’t have the bandwidth to watch every rep, correct every movement, or catch every mistake before it turned into inefficiency — or injury.

We all experienced moments where poor form held us back. Sometimes it meant missed PRs. Sometimes it meant getting hurt. And almost always, it meant slower progress than we knew we were capable of.

So we started asking ourselves: What if high-quality coaching was accessible anytime?
What if athletes could get reliable feedback even when coaches weren’t physically present?
And what if form correction and technique analysis could be delivered instantly — using technology?

That question became the seed for Persis.

What It Does

Persis uses computer vision and AI to give athletes real-time form correction, automatic workout logging, and long-term performance analytics.
Athletes train smarter with immediate feedback, while coaches get data they can use to support athletes better — even remotely.

Ultimately, Persis brings elite-level coaching insights to anyone with a camera.

How We Built It

Backend built with MongoDB for scalable data storage.
Web platform built using TypeScript, Node.js, and React for a fast, modern interface.
Real-time movement analysis powered by X AI models streamed through WebSockets.
A standard webcam captures athlete movement directly through the browser.

This pipeline allows video → pose detection → feedback to happen almost instantly.

Challenges We Ran Into

Difficulty using X AI to analyze live footage without jitter or instability.
Training the computer vision model to recognize complex motions like rowing and squatting.
Issues with skeleton jitter, noisy keypoints, and unstable model predictions.
Backend–database communication challenges when syncing real-time CV output.

Each challenge pushed us to refine our engineering and deliver a more polished system.

Accomplishments We're Proud Of

Successfully training a CV model to recognize complex, multi-joint athletic movements.
Achieving accurate capture of rowing mechanics — one of the most technically challenging motions in sports.
Building a working full-stack system that delivers real-time feedback through a webcam.

What We Learned

How to integrate hardware and software into one seamless experience.
How to train, fine-tune, and deploy computer vision models.
How to build a full-stack architecture with live AI inference.
How to turn a personal problem we faced as athletes into a real product with real impact.

What's Next for Persis

Sport-specific modules for rowing, sprinting, hurdling, golf, and more.
A fully integrated coach–athlete ecosystem with messaging, training programs, and athlete management tools.
Advanced analytics like fatigue detection, movement asymmetry, and technique progression.
Mobile app integration for seamless, on-the-go training.
Team and club dashboards for schools, universities, and performance gyms.

Our vision is to make Persis the new standard for AI-assisted athletic development — accessible, intelligent, and built for every athlete striving to get better.