Your Fitbit Air is getting a whole lot better at being able to recognize all your exercise activity

Google has recently announced significant upgrades to the Fitbit Air's ability to automatically recognize different types of exercise activities. By combining sensor data with location and routines, Google is employing an advanced language model (LLM) to improve activity classification. This upgrade is expected to result in a remarkable 200% increase in accuracy and a reduction in missed workouts by half.

Screenless activity trackers like the WHOOP and the new Fitbit Air have gained traction among users who value their simplicity and long battery lives. However, these devices face unique challenges. Without a screen to provide feedback or an easy way to interact with the hardware, automatic detection of various workouts becomes crucial.

Google has taken notice of this issue, as it is the top feature request from Fitbit Air users. In response, the company has developed a new motion detection software model for the wearable. This enhanced software will not only improve detection for activities like swimming or weight training but also leverage LLM reasoning to integrate factors such as the wearer's location and routine when classifying activity.