Google’s annual I/O developer conference wrapped its keynote on May 19, with the company once again building almost the entire show around its Gemini AI effort. CEO Sundar Pichai framed 2026 as a year of “hyper progress,” while also conceding that this is the stage where users want to see AI translate into day-to-day usefulness. Google says 900 million people now use the Gemini assistant and that more than 50 billion images have been generated with it.
Here are the eight announcements worth knowing about.
1.Two New Gemini Models: 3.5 Pro and 3.5 Flash
The headline release is Gemini 3.5, available today in both Search and the Gemini app. Alongside it, Google launched Gemini 3.5 Flash, a cheaper, lighter variant aimed at users who don’t need the full Pro model. Both went live immediately.
2. Google Search Becomes Agentic
Google has folded its AI agents directly into Search through what it calls the “intelligent search box,” rolling out to everyone starting today. Beyond text answers, the box can generate explanatory images or short video clips on the fly — Google’s onstage example produced a video walking through how a black hole forms.
A companion feature, Generative UI, builds custom layouts for search results in the browser depending on whether the most relevant response is a video, an image, or an article. Generative UI is slated to reach all users over the summer.
3. A Redesigned Gemini App, Plus Daily Brief and Ask YouTube
The Gemini app gets a visual overhaul Google calls Neural Expressive, with new backgrounds, a new typeface, fresh voice-chat animations, and a wider range of regional voice accents.
Two notable additions sit alongside it. Daily Brief is a personalized morning digest that pulls from your calendar, email and other data to summarise and prioritise your day. Ask YouTube lets you pose natural-language questions on the platform, how to grill a steak, for example, and surfaces relevant clips, jumping straight to the timestamp that answers the query. A separate feature, Docs Live, lets you draft documents by voice, with an agent generating text and pulling in web citations as you talk.
It’s worth noting Google has flagged that these features may eventually carry ads.
4. Gemini Omni, A Sora-Style Video Generator
Google announced Gemini Omni, a video generator positioned against OpenAI’s now-discontinued Sora 2. Omni leans heavily on putting the user’s own face front and center. For instance, selfie videos can be restyled with different backgrounds and environments, and footage from your camera roll can be turned into various cinematic styles.
Google says still-image and text generation will follow later. A starter version, OmniFlash, is available today for Google AI+ Pro and Ultra subscribers.
5. Gemini Spark, An Always-On Personal Agent
Spark is Google’s answer to the viral assistant OpenClaw. It’s pitched as a personal agent that learns the rhythms of your schedule, handles recurring tasks, drafts emails, and pulls information from your Google Drive files. Because it runs on Google Cloud rather than your device, Spark can process tasks in the background without your hardware staying on. For now it works only with Google’s own software, Chrome and third-party support are promised for later in the summer.
6. Agentic Shopping With a Universal Cart
Google is rolling out an agentic shopping experience built around a single universal cart that spans multiple retailers. As you browse, the agent organizes a wish list, flags price drops, and notes new versions or color options. You can ask Gemini for product details or to hunt for better deals, then check out across several stores at once using Google’s payment system, or opt to buy from the original retailer directly.
7. Flow Gets New Creative Tools
Google’s Flow toolset, aimed at people without camera gear or editing skills, picked up new capabilities for making promo videos, invitations, music videos and short films. One example: upload a single photo and Flow generates 16 short clips that animate the scene while preserving the people and setting. Musically, a rough melody sketch can be built into a fuller produced track in a chosen genre.
8. Google’s “Intelligent Eyewear” Arrives This Fall
Google is explicitly avoiding the term “smart glasses.” There are two categories of its Android XR “intelligent eyewear”: an audio-only design and a version with a small in-lens display. The audio glasses ship first, this fall, made with Warby Parker and Gentle Monster, with speakers in the temples and cameras that let Gemini see what you’re looking at. Pricing is still to come.
The display versions remain in development and will follow later; they can show messages, directions and search results, and overlay live translation in your field of view. Both types tie into Gemini voice chat and Google’s Nano Banana image tool, and the project is a partnership with Samsung on the underlying tech.
