TrueForm

ENGINEERING THE
FUTURE OF HUMAN
PERFORMANCE.

TrueForm is a personalised AI workout assistant: computer vision, sequential deep networks, and ensemble models work together for real-time pose correction and workout classification—lifting performance while reducing long-term injury risk.

Research-grade signal, pocket-sized

We leverage Google MediaPipe for real-time pose estimation, then train low-latency PyTorch heads: a Soft Actor-Critic policy for twelve joint corrections (~5% MAE), a deep net for twenty-two workout classes (98% F1), and a Random Forest over seven muscle-group activations (90% F1). A modular CI/CD pipeline keeps those models shipping safely; we are actively exploring rehabilitation and physiotherapy extensions on the same stack.

98% Workout macro F1
<1ms Model inference
bolt

SPEED OF INSIGHT

MediaPipe landmarks feed PyTorch SAC, DNN, and Random Forest heads with sub-millisecond inference on deployed regressors and classifiers.

Biometric Safety

Sequential control plus ensemble classification flag risky mechanics early; rehab and physio scenarios are on the roadmap.

Marcus Thorne

Founder & CEO

THE HUMAN BODY IS THE
ULTIMATE INSTRUMENT.

"We've spent the last decade perfecting the hardware of sport—shoes, sensors, and fabrics. But we've neglected the most critical variable: the software of human movement."

Marcus Thorne founded TrueForm with a singular mission: give every athlete the same analytical oversight once locked inside pro labs. Today that vision is expressed through MediaPipe-driven pose streams, reinforcement-learning joint control, and ensemble workout intelligence—all continuously refined through disciplined ML operations.

verified Read the Kinetic Whitepaper

SECURE YOUR
EARLY ACCESS.

For enterprise partnerships, coaching integration, or priority HUD access, submit your protocol request below.

hub

Global Network

Join 200+ elite training facilities using TrueForm.

security

Data Sovereignty

End-to-end encrypted biometric storage.