Fitbit To Use Google Gemini to Introduce AI-Powered Health Features

During Google’s annual Check Up event on Tuesday, it was announced that Fitbit would soon incorporate health-focused artificial intelligence (AI) features, leveraging Google’s Gemini platform. Rather than introducing features for specific tasks within particular workflows, Google Research will collaborate with Fitbit to develop a large language model (LLM) tailored to personal health.

VP of Engineering and Research at Google Yossi Matias revealed in a blog post about the collaboration between Google Research and Fitbit. Together, they are developing a Personal Health Large Language Model aimed at enhancing personalized wellness and health features within the Fitbit mobile app. This model is designed to provide users with improved recommendations and insights directly from their Pixel and Fitbit devices.

As an illustration, Google highlights its capability to analyze sleep patterns and offer recommendations for adjusting exercise routines based on these insights. This aligns with Google’s presentation of Fitbit Labs during its Made by Google event last year, where it showcased how its AI could analyze data from recent running workouts to provide insights into why today’s run may have felt more challenging than previous ones. 

Similar to many health features found on contemporary fitness trackers and smartwatches, Google’s Personal Health Large Language Model does not intend to provide medical diagnoses or treat any conditions. Instead, the model aims to simplify the navigation and understanding of Fitbit data.

The AI model is currently undergoing testing and refinement to ensure optimal performance, although Google has not provided a specific timeline for the release of these features.

Matias outlined the anticipated capabilities of the Gemini-backed LLM for Fitbit, including actionable messages and personalized coaching based on user data from Google-branded wearable devices and Google accounts. For instance, the AI could analyze sleep quality and patterns to offer workout intensity recommendations.

The blog post further disclosed that the Gemini-based AI model is being trained on diverse sets of health signals from de-identified research case studies, in collaboration with wellness experts and accredited coaches. 

Premium Fitbit users will have the opportunity to preview these features before their public release through Fitbit Labs, a platform similar to Google’s AI Labs, where users can register to test and provide feedback on features in development.

We anticipate gaining further insights into the company’s plans for generative AI at its annual developers’ conference known for providing important updates on its technologies and products.