TL;DR

• •Google has some big new upgrades for Fitbit Air’s ability to automatically detect workouts.
• •Combining sensor data with location and routines, Google’s using an LLM to better classify activity.
• •That should result in 200% more accurate classification, and half as many missed workouts.

Screenless activity trackers like WHOOP and the new Fitbit Air have really been taking off with users who appreciate their simplicity and long battery lives, but devices like these present special problems. When we don’t have a screen to provide feedback and an easy way to interact with the hardware, being able to do lots of stuff automatically is incredibly important. Today we’re learning about a big new upgrade for the Fitbit Air in this direction, as Google beefs up its ability to automatically detect different workouts.

Google reveals that improving workout detection has been the biggest feature ask it’s heard from Air owners. And in order to give them that, engineers went back to the drawing board and wrote up a whole new motion detection software model for the wearable. In addition to offering better detection for things like swimming or weight training, it’s also now leaning on LLM reasoning to combine factors like your location and routine with movement data in order to classify activity.