A Samsung Galaxy S25 Ultra smartphone during a media event ahead of Galaxy Unpacked in New York, US, … [+]
Alongside news that Gemini has replaced Bixby on Samsung phones this week, Google has updated Gemini’s abilities on millions of Android phones.
Gemini can now complete tasks across several apps with a single prompt. For example, I asked the AI service to find Caribbean restaurants in North London and send them to my friend on WhatsApp, which it did. I also asked it to find out when Mobb Deep are next playing in London and add it to my calendar, which it completed.
This new multimodal feature works in conjunction with Gemini’s app extensions. A handful of Google apps (and now select Samsung apps) feed into the AI’s responses, including Workspace apps, Spotify, Messages, and WhatsApp.
This is an important step in making AI assistants more like actual assistants. What has lacked throughout the Alexa, Siri and Google Assistant era is the seamlessness of connecting to other services and completing simple tasks. You would imagine asking a Google Nest Speaker to add ingredients to a shopping list is easy, but it is that type of functionality has escaped these products for years.
This has meant voice-activated assistants haven’t lived up to the promise or hype. The multimodal skill adds a layer of intelligence to a patchy experience, but it is still limited to select apps and it can be confusing what the new feature’s limitations are.
The key to using the feature is to remember that it only works with select Google and Samsung apps. Gemini found my housework task list in Google Documents and shared it with my partner, as requested. But it won’t send a message on your behalf, so I couldn’t get it to text my barber to book an appointment and then add that my calendar.
Also, the example Google gives of how the automation works, asking it for restaurant recommendations and then sending the list to a friend, removes a key part of using Google Maps: quality control. I want to vet establishments before I send them on to someone, which still needs a human touch.
Spreading Gemini far and wide
Elsewhere, Gemini Live—the conversational part of the AI service that allows instant back-and-forth chat—now supports uploaded images, files, and YouTube videos. For example, you might upload a picture of your fridge contents during a discussion with Gemini Live about what recipes to cook that week. This is only available on Pixel 9, Galaxy S24 and S25 phones.
Circle to Search has also been updated with AI Overviews. The search function, which launched last year, allows users to draw a circle on images to conduct an image search. Now, Google’s AI summaries that appear on top of Google searches will also show up in Circle to Search results, although the accuracy of overviews has been criticized since it launched last May.
With the exception of the Gemini Live updates, it is interesting that Google is adding next-generation AI experiences to millions of smartphones without necessitating new hardware or a subscription.
Google says the new multimodal Gemini feature is available to all Gemini users on the web, Android and iOS. Google does have a paid AI subscription tier and Samsung has teased charging for Galaxy AI at the end of this year. For now, though, these AI updates are free and accessible to most smartphone users.