Wear OS Weekly
My new weekly column focuses on the state of Wear OS, from new developments and updates to the latest apps and features we want to highlight.
Gemini is likely coming to Wear OS watches, but the details are vague at best. We don’t know when it’ll arrive, whether tools like Gemini Live and app extensions will work, how Gemini’s answers will look on tiny watch screens, and whether all the processing will require your phone on hand.
With the right answers to these uncertainties, Google could give Wear OS users a real excitement boost with Gemini smarts. Major Pixel 9 sales growth proves that Gemini marketing is working on consumers, and getting AI insights with your phone in your pocket will appeal to many.
But Google Gemini isn’t just one service: It’s a label applied to different functions and levels of intelligence, varied based on whether it’s cloud- or device-based. Wear OS’s version of the assistant/chatbot would significantly differ from what your Pixel 9 Pro or laptop offers, for example.
So, what will Gemini on Wear OS do differently than Assistant? Or, at least, what do I hope Gemini will do differently, and can Google make Gemini work on such limited hardware? Let’s dive in.
What little we know about Gemini on Wear OS
9to5Google discovered the Gemini references in Google app beta code, with the language “Easily talk back and forth to get more done with an assistant on your watch, reimagined with Google AI” describing the wearable app.
The site also found references to Gemini being summonable by the same methods as Assistant: Using the “Hey Google” phrase or holding down the Pixel Watch 3‘s side button. You can safely assume that Google will keep its always-listening option for Gemini, too, though this will burn through battery life faster.
Google confirmed in November that the Pixel Watch series’ next Wear OS update would be in March. Wear OS updates usually arrive monthly with minor changes, so why is the Watch 3 stuck with its September launch software? Perhaps because Google needed six months to get Gemini ready in wearable form, pushing other updates to the back burner.
If not March, we could see Gemini arrive alongside the upcoming Samsung Galaxy Watch 8 or Pixel Watch 4, both of which should launch with Wear OS 6 in late summer. Google in particular will probably have new SoC hardware by then; Qualcomm’s VP of wearables Dino Bekis told me their 2025 Wear chip will have “more on-device AI capability.”
Gemini Nano, the portable version of Google’s AI, still requires a phone’s neural processing unit for edge tasks like summarizing an email or transcribing conversations. So it’s likely that some Gemini on-watch tools will be exclusive to newer models that add NPUs, while older Wear OS watches with decade-old Cortex cores will have to rely solely on your phone and cloud computing.
What Gemini could bring to Wear OS watches
The most obvious Gemini Wear OS application is to enable on-wrist conversations, like you get with Gemini Live, or simply more intelligent answers to questions. The leaked language “talk back and forth” sure seems to imply an actual conversation, which gives me hope that you’ll be able to have in-depth discussions with remembered context until you have the answer you need.
The biggest hurdle is that most Gemini answers take about 10–15 seconds if they require pulling up information — compared to 4-ish seconds for simpler questions — and phone-to-watch Bluetooth communication may add an additional delay. Google Assistant has gotten faster at on-device answers, but Gemini will feel like a step backward for speed — even if it’s a step forward for clarity.
The second hurdle is the size of the screen. Gemini spits out a ton of information to any question or prompt to provide enough context (see the photo above). Ask your Wear OS watch the same question, and you’ll have to scroll through a lot more tiny text or stand and wait for Gemini to read everything out loud for a couple of minutes.
The next potential Gemini application on Wear OS is to work with Gemini extensions. Imagine being able to prompt Gemini from your wrist to summarize your unread emails, create a new YouTube Music playlist based on your mood, or tell you about nearby 4-star lunch spots via Maps. Your phone would do all the work, but it would remain in your pocket.
Google began integrating Gemini and Google Home last year, and it’s hard to say that Gemini is any better than Assistant was at controlling your smart home tech, barring a few minor perks like summarizing who came to your door via Nest Doorbell. But any future innovations will be Gemini-exclusive, so Wear OS owners with smart homes will benefit in the long run.
Then, of course, you have Gems, or the customizable and specialized versions of Gemini built for specific tasks like playing chess or serving as a “running coach,” one Google example that feels especially apt for Wear OS.
I think specialized Gems you can quick-swap between in a Gemini Tile would be perfect for the format, because no one wants to stand around waiting for generalized answers on your wrist. If you know what topic you’re usually curious about while on the go, why not have that Gem quick-loaded on your Pixel Watch?
Beyond these obvious Gemini applications on Wear OS, I’d be interested to see if the app’s image generation tools could be put to use for watch face creation. Imagine if you could use Imagen 3 to create an XML-compatible face cropped to your screen’s proportions in less than a minute.
Gemini could also prove useful for finding new Wear OS apps, which can be a pain to find in the on-watch Play Store. You could ask the assistant to recommend a few options for different categories (e.g., fitness, weather, productivity, etc.) and summarize their differences and strengths.
The most obvious Gemini application is health and fitness
If it’s not apparent from the list above, I’m hopeful about Gemini on Wear OS, but also skeptical that Google can deliver a transformative Gemini assistant experience with lightweight hardware and a tiny screen. I’m ready to be proved wrong, but I don’t know if smartwatches can support enough tokens to provide the context necessary for intelligent answers. So I’m expecting delayed answers and a significant battery drain while using Gemini for long conversations. We’ll find out soon!
I could see Gemini proving more useful in the backend. If Google uses Gemini to analyze your Fitbit results upon request — ideally while preserving your data privacy — that could provide useful insights on your current health, recent fitness or sleep trends, and so on. It could then highlight the key AI insights in your Morning Brief.
On the fitness side, Google already recommends daily runs to Pixel Watch 3 users based on their training load. In theory, a Gemini Coach could take things to the next level and analyze post-run results (like Strava’s Athlete Intelligence) or expand beyond running and recommend gym workout routines, for example.
Google could even provide those insights during an activity; you could say, “Hey Google, how am I doing?” and have it tell you that you’re on PR pace or that your heart rate is too elevated over your wireless earbuds.”