Note: This post reflects the application as of January 2026 and should be considered historical information beyond version 6.2.
We’re pleased to announce the release of version 6.2 of our application. This minor update introduces several new features and improvements.

Users with a keen eye may have noticed the off-piste toggle in the settings screen over the last few releases. This reflected an experiment to extend our indoor tracking model to outdoor environments. While it produced usable results, mountain conditions differ significantly from the controlled environment of an indoor ski slope.
Improving this required data, and a substantial amount of it. Much of this year’s ski trip was spent recording and labelling gigabytes of training data. Drawing on our experience building the indoor tracker, we were able to fine-tune a model specifically for outdoor use and reach a rough and ready 91.6% accuracy. Alongside this, we introduced outdoor-specific filtering to analyze the model's output into a complete ski session. This enables Piste to be used in resorts we do not yet support, as well as on off-piste sections within supported resorts.
The feature remains opt-in. To enable it, open the settings screen and turn on the toggle. Your device must also include a barometer, which is available on most iPhones, Pixel devices, and Samsung S-series phones.
We do not use large language models or conversational systems. Instead, we rely on an on-device machine learning approach known as a classifier. This processes short windows of recent sensor data from your phone and assigns the activity category that best matches the observed patterns.
Many GPS watches and tracking apps rely primarily on altitude gradient. In our testing, this proved unreliable. Uphill sections of pistes, or gondolas passing between ridgelines, frequently distort the altitude readings and lead to incorrect tracking.
If you're like us, we often forget to stop recording and may walk back to accommodation with tracking still enabled. These segments should not be recorded as ski runs. The classifier has been trained to distinguish between skiing, walking, and lift usage based on sensor behaviour. It evaluates current data in the context of recent activity and decides what should be recorded.
At present, no user data is used for training. All training data has been collected by our own team while skiing in La Plagne and on indoor slopes.
There are, however, many skiing and snowboarding styles. In future releases, users will be able to opt in to contribute their data to improve detection accuracy across a broader range of conditions and techniques.
For best results, store your device in a pocket or bag away from direct body contact, such as in a ski jacket rather than trousers. Avoid placing pressure on the device whilst skiing, as this can affect barometric readings and lead to incorrect altitude measurements. The effect is similar to squeezing a balloon and changing the internal pressure.

Our recent trip exposed several weaknesses in lift detection. Most of these have been addressed in this update. The system is now more tolerant of gps inconsistencies, and valid data can now recover a degraded detection score. Previously, scores could only decline, which meant longer lifts were often rejected.

We have added Bardonecchia, a user-requested resort and a former Olympic venue.
We have also added Whistler, Canada 🍁 as our first North American resort, released as a beta. Some localisation work remains, particularly around North American run grading, but this marks an initial step towards broader coverage in the US and Canada.
If your local ski area is not listed, feel free to email us and we will review it for inclusion
Piste is now tracking the time spent stationary on runs, a UI update will expose this data so you know how long you were waiting for Jerry to catch up!
We are looking at adding a spreadsheet export of your day; for the real number crunching nerds.
We are also looking at improving our AI model with even more crazy maths
Interested in the update? Download the latest version on the app stores: