Google used Tuesday’s keynote at its annual Google I/O developers conference near its headquarters in Mountain View, California, to lay out a broad push toward what it describes as “agentic” artificial intelligence, alongside new Gemini models and upgrades to products such as Google Search and Chrome.
Pichai set the tone by telling developers that Google is “firmly in our agentic Gemini era.” He said the potential of agents is visible but also described the technology as early, particularly in making agents “easy to use, super secure and truly helpful.” Google and its parent company Alphabet have invested heavily in artificial intelligence, and Pichai referenced both ongoing spending and the company’s recent business momentum.
A central part of Google’s announcements was the rollout plan for the next Gemini model line. Google said its Gemini 3.5 models are beginning to roll out Tuesday to billions of global users, starting with Gemini 3.5 Flash. The Flash model, according to Google, is designed to prioritize speed and is also positioned as the strongest agentic and coding model yet, and the company said Flash will become the default model in the Gemini app and in “AI mode” on Google search.
Google also said it is working on Gemini Pro 3.5, which it plans to launch next month. Alongside the new models, the company said Gemini 3.5 has been developed with updated safety training and mitigations, aimed at reducing the chance that models generate harmful content or mistakenly refuse to answer safe queries.
In addition to new text and coding capabilities, Google previewed Gemini Omni, a model family described as enabling users to create high-quality video by entering prompts with multiple types of inputs. Google said users will be able to provide any combination of text, images, videos, and audio in a query, and then edit the generated video through conversation with the model. Google did not provide specific timing for when users would be able to create images and audio with Omni, but it said it expects Omni-generated videos to appear more realistic because of how the system understands physical dynamics such as gravity, kinetic energy and fluid dynamics.
The first Gemini Omni model, Gemini Omni Flash, is scheduled to launch Tuesday for Google’s AI Plus, Pro and Ultra subscribers in the Gemini app and Google Flow. Google said the model would also become available at no cost on YouTube Shorts and the YouTube Create App beginning this week, and it said Omni-generated videos will include an imperceptible digital watermark using SynthID. Separately, Google said it is adding content credentials verification to the Gemini app that can help determine whether certain content such as photos or video was created with artificial intelligence or captured with a phone camera and then edited with AI tools; it said the feature will appear in search in Chrome in the coming months and that OpenAI, Kakao and Eleven Labs are adopting SynthID technology for more of their AI-generated content.
Google also gave developers a preview of Gemini Spark, describing it as a proactive agent powered by Gemini 3.5. The company said Spark will be able to handle routine tasks such as sorting through meeting notes, emails and chats, then producing a document that includes the biggest takeaways and to-dos. Unlike some agents, Google said Spark is based in the cloud so it can continue working even after users shut their laptops or lock their phones.
The company said Spark is designed to ask for permission before carrying out “high-stakes” tasks, including sending an email or making a purchase, addressing concerns that agents could otherwise act without sufficient user control. Google said select testers would get access beginning Tuesday, and that it plans to roll out a beta mode to U.S.-based subscribers of its Google AI Ultra tier. It also said Spark will operate directly within Chrome later this summer.
At Google I/O, the company additionally discussed updates to search and shopping. It described “AI mode” in Google search as giving users more conversational answers before providing relevant links, and said those queries have more than doubled every quarter since the feature began rolling out last year; Google also said the tool surpassed 1 billion monthly users, citing Liz Reid, head of search. For the default experience, Google said the new search model in AI mode will be Gemini 3.5 Flash and it introduced what it called an intelligent search box.
Google said the updated search box will adapt to longer queries and can help users write out questions using AI-powered suggestions rather than traditional autocomplete. The company also said users can search using multiple modalities—text, images, video, files and even Chrome tabs as search inputs—and that the search box rollout is starting Tuesday in all countries and languages where AI mode is currently available.
For shopping, Google announced a new tool called the Universal Cart, which it described as an “intelligent shopping cart” that can work across merchants and across services. Google said users can add items to a cart while browsing Google search, chatting with Gemini, watching YouTube, or reading emails in Gmail, and that the cart will then use Gemini models to run tasks such as looking for deals and price drops, providing price history information and alerting users when items return in stock. Google said Universal Cart will be available to users on search and the Gemini app this summer, with YouTube and Gmail to follow.
Separately from Gemini models and search updates, Google previewed smart glasses in two categories. It said audio glasses that provide spoken help in the ear are expected to arrive later this fall, while display glasses would be part of the broader lineup. Google said users will be able to say “Hey Google” or tap the side of the frame to access Gemini, which will then provide assistance for navigation, managing communication on a phone, real-time translations and other tasks.
Google said it partnered with Samsung and eyewear brands Gentle Monster and Warby Parker to develop the glasses. The company showed two designs Tuesday—sunglasses from Gentle Monster and glasses from Warby Parker—and said those designs will launch as part of the brands’ full collections later this year.