Google is quietly giving the Gemini app a much bigger productivity push ahead of I/O 2026. Some users are now spotting a new “Extended” thinking mode alongside a wave of upcoming third-party app integrations.
The new feature appears inside Gemini’s model picker and adds a “Thinking level” option when using Gemini 3 Flash or Gemini 3.1 Pro. Users can currently choose between Standard and Extended. However, the rollout seems fairly limited for now.
While Google hasn’t officially explained what “Extended” does, it looks similar to the adjustable reasoning modes already available in Google AI Studio. There, users can increase how deeply Gemini processes prompts before responding. In practice, that could mean slower but more detailed answers for things like planning, research, or complex creative tasks.
But the bigger update might actually be the app integrations Google is preparing behind the scenes.
Gemini already works with apps like GitHub, Spotify and WhatsApp. Support documents reveal that integrations for Canva, Instacart and OpenTable are also on the way.
The Canva integration looks especially ambitious. Gemini will reportedly be able to create designs from prompts, move files between folders, manage comments, and even generate images directly into Canva projects. One example prompt shown asks Gemini to create a rustic wedding invitation with a dusty rose and sage green colour palette.
Instacart support appears more practical, letting users add groceries directly to carts through Gemini prompts. That includes building ingredient lists from recipes, checking nearby store availability, or creating shopping lists conversationally.
Meanwhile, OpenTable integration could turn Gemini into more of a digital concierge. Support documents suggest users will be able to search for restaurants, book tables, cancel reservations, and even add dinner plans to Google Calendar. All of this can be done without leaving the app.
Taken together, the update shows Google pushing Gemini beyond being just a chatbot. The company seems increasingly focused on turning it into a central AI assistant. In fact, it could actually complete tasks across multiple services. It would not just answer questions about them.
Google hasn’t announced a wider rollout date yet. However, with I/O 2026 just around the corner, it feels unlikely these features will stay hidden for long.
