“Visible format generates an immersive, magazine-style view full with images and modules,” says Josh Woodward, VP of Google Labs, Gemini, and AI Studio. “These components don’t simply look good however invite your enter to additional tailor the outcomes.”
With Gemini 3, Google can also be introducing Gemini Agent, an experimental characteristic designed to deal with multi-step duties immediately contained in the app. The agent can connect with providers reminiscent of Google Calendar, Gmail, and Reminders. As soon as granted entry, it may possibly execute duties like organizing an inbox or managing schedules.
Just like different brokers, it breaks duties into discrete steps, shows its progress in actual time, and pauses for approval from the person earlier than persevering with. Google describes the characteristic as a step towards “a real generalist agent.” It will likely be obtainable on the net for Google AI Extremely subscribers within the US beginning November 18.
The general method can appear lots like “vibe coding,” the place customers describe an finish aim in plain language and let the mannequin assemble the interface or code wanted to get there.
The replace additionally ties Gemini extra deeply into Google’s current merchandise. In Search, a restricted group of Google AI Professional and Extremely subscribers can now swap to Gemini 3 Professional, the reasoning variation of the brand new mannequin, to obtain deeper, extra thorough AI-generated summaries that depend on the mannequin’s reasoning quite than the prevailing AI Mode.
