We make AI movement
accessible to every developer
QuickPose was built by a team who spent years wrestling with the complexity of pose estimation — and decided there had to be a better way. Now we give that capability to developers in hours, not months.
From a tennis court to the world's movement stack
It started in 2019 with a problem in tennis. We founded Sliced Backhand Ltd to build AI-powered pose estimation tools for tennis coaching — analysing technique, tracking movement, and helping players improve their game using just a smartphone camera.
Then COVID hit. Gyms and courts closed overnight, and we pivoted fast — building a reflex training app that people could use at home. The app grew to 100,000 downloads and found an audience we never anticipated: physiotherapists and rehab clinicians using it to support ACL recovery and stroke rehabilitation patients.
That experience changed everything. We'd seen first-hand how powerful body tracking could be — not just for sports, but for health, for rehabilitation, for human wellbeing. And we'd also felt every sharp edge of building it ourselves: the model tuning, the performance trade-offs, the platform differences, the integration complexity.
All of that hard-won knowledge went into QuickPose. Launched in January 2023, it distils years of real-world pose estimation experience into an SDK that any developer can drop into their app in an afternoon.
How we got here
Started building AI pose estimation tools for tennis coaching — our first deep dive into body tracking on mobile devices.
With courts and gyms closed, we pivoted to a home-based reflex training app. It reached 100,000 downloads and was adopted for ACL and stroke rehabilitation — expanding our sense of what pose estimation could do.
We decided to stop rebuilding the same infrastructure across every project. Every lesson from Sliced Backhand went into designing a developer SDK — one that handles the hard parts so builders can focus on their product.
The iOS SDK shipped — giving developers a production-ready pose estimation engine for fitness, yoga, sports biomechanics, and health apps. Built on MediaPipe's open-source foundation.
QuickPose expanded to Android — bringing the same production-ready pose estimation to the full mobile ecosystem, with native Kotlin and Jetpack Compose support.
QuickPose passed one million sessions — a milestone that represents millions of real people getting form feedback, tracking their recovery, improving their movement, and pushing their performance through apps powered by our SDK.
Principles of our work
We've built enough software to know what matters — and what gets in the way.
Developer experience first
A powerful SDK that's painful to use isn't powerful. We obsess over documentation, sample projects, and SDK design — because your time spent integrating is time not spent on your product.
Privacy by architecture
On-device processing isn't a selling point we added — it's how the system was designed from the start. No video leaves the device. No data is ours to monetise.
Real-world accuracy
We tune for how people actually move — in imperfect lighting, at odd angles, on mid-range phones. Lab accuracy is easy. Production accuracy is the job.
Open foundations
QuickPose is built on MediaPipe — Google's open-source ML framework — so you're never locked into a black box. Transparent, auditable, and free to inspect.
Built on MediaPipe
QuickPose is built on top of MediaPipe, Google's open-source framework for on-device ML solutions. That means the pose estimation foundation you're building on is transparent, battle-tested, and maintained by one of the world's leading ML teams.
⚖️Let's Explore What You Can Build
Whether you're integrating the SDK yourself or want a team to build your AI movement feature end-to-end — we'd love to hear about your project.
Book a Discovery Call →Or email us at info@quickpose.ai
