About Bikendure

Bikendure was born out of a simple frustration: professional bike fitting is expensive, geographically limited, and usually requires a studio appointment days in advance. We believed there was a better way.

By combining computer vision, biomechanical research, and large language models, we built a tool that can analyse your riding position from a short video — delivering results that match a manual studio fit, at a fraction of the cost.

Our mission

To make performance cycling accessible to everyone — from weekend riders to competitive athletes — by putting professional-grade tools in their hands.

The technology

We use MediaPipe’s pose detection model to identify 33 body landmarks on every frame of your video. Our analysis engine then calculates joint angles throughout the full pedal stroke and compares them against evidence-based optimal ranges from sports science literature.

Results are interpreted by a GPT-4 powered AI that translates raw measurements into actionable English (and soon, multiple languages) — no biomechanics degree required.

Get in touch

We’d love to hear from you. Reach out through our contact page.