SAN FRANCISCO, May 19 (AP) Google will soon announce a wealth of new tools and systems powered by artificial intelligence, including an AI assistant that actively performs tasks on your behalf and assists you.
“Agentic” AI, the latest buzzword for tech companies, was a central focus at Google’s annual developer conference, Google I/O. Gemini Spark, its next AI agent, was one of the company’s many announcements at Tuesday’s conference.
“We are firmly in the Gemini agent era,” Google CEO Sundar Pichai said Tuesday in front of a packed outdoor theater near the company’s Mountain View, Calif., headquarters. “We’ve tried all kinds of agents, and we really see the potential. But we’re still in the early stages of making agents easy to use, very secure, and really useful.” Google and its parent company, Alphabet Inc., have poured billions of dollars into AI development. The company’s finance chief said on a conference call with investors in late April that capital spending this year could reach $190 billion. However, the investment appears to be paying off, with quarterly profits showing strong growth. Since the news, the stock price has increased another 11%.
Pichai said in his keynote that the Gemini app had 400 million monthly active users last year and now has more than 900 million users, more than doubling in one year.
Click here for the latest version of Gemini
Google’s newest family of models, Gemini 3.5, will roll out to billions of users around the world starting with Gemini 3.5 Flash on Tuesday. The Flash model is focused on speed, and Google says 3.5 Flash is its most powerful agent and coding model to date, while being approximately four times faster than some of its competitors.
This model is now the default for the Gemini app and Google Search’s “AI mode.” The company is also working on a 3.5 version of Gemini Pro, which it uses internally and expects to be released next month.
Gemini 3.5 has been developed with new and more advanced safety training and mitigation measures to make its models less likely to generate harmful content or falsely refuse to answer safe queries, the company said.
Google also announced a new model Gemini Omni. This allows users to create high-quality videos by querying any input, including text, images, video, and audio. Videos created by Omni can be easily edited through conversations with the model. Eventually, users will be able to create images and sounds with Omni, but details about when those features will be rolled out have not been disclosed.
The company said Omni’s videos look more realistic than videos created with other models because it understands forces such as gravity, kinetic energy, and fluid mechanics.
Gemini Omni Flash, the first product in the Omni family, launches Tuesday for Google Al Plus, Pro, and Ultra subscribers through the Gemini app and Google Flow. Starting this week, it will be available for free on YouTube Shorts and the YouTube Create app.
All videos created with Omni include SynthID, Google’s imperceptible digital watermark, but Google also adds content credential validation to the Gemini app. The tool determines whether content, such as a photo or video, was created by AI or was taken with a phone’s camera and edited with an AI tool. It will be available in Chrome Search in the coming months. Google also announced that AI companies Open AI, Kakao, and Celebrities are using its SynthID technology for much of their AI-generated content.
24/7 agent
Gemini Spark, powered by Gemini 3.5, lets you organize meeting notes, emails, chats, and complete everyday tasks like creating documents with your biggest takeaways and to-dos. Unlike other available agents, Spark is cloud-based, so it continues to work in the background even when users close their laptops or lock their phones.
The proactive nature of AI agents is what sets them apart from chatbots, and it also contributes to concerns about the power of the technology. Gemini Spark is designed to ask for permission before performing “high-stakes” tasks such as sending emails or making purchases, the company said.
Some testers will have access to the agent starting Tuesday, and the company plans to roll out beta mode to U.S.-based Google AI Ultra tier subscribers.
Later this summer, Gemini Spark will work directly within Chrome, the company says.
The rise of AI in search and shopping
The most talked about development at last year’s conference was the introduction and rollout of “AI mode” to Google’s search engine. This feature builds on previously implemented changes to user platform experiences and interactions to provide more conversational answers to user queries before providing relevant links.
Google’s head of search, Liz Reid, said AI mode queries have more than doubled every quarter since its launch last year, and the tool recently surpassed 1 billion monthly users.
The new default model for search is Gemini 3.5 Flash, and the company is introducing what it calls an intelligent search box. Reid said the change is the biggest upgrade to the search box in 25 years and means the box can now accommodate longer queries and allow users to write out questions with AI-powered suggestions rather than traditional autocomplete.
Users can also search using multiple modalities, including text, images, videos, files, and even using Chrome tabs as search inputs. The new search box will begin rolling out Tuesday in all countries and languages where AI mode is currently available.
The company also announced a new tool, Universal Cart, which it calls a “truly intelligent shopping cart.” This works across merchants and services, so users can add items to their cart while browsing Google Search, chatting with Gemini, watching YouTube, or reading emails in Gmail. The cart then runs on the Gemini model and works as soon as an item is added to the cart, looking for deals and price drops, providing price history information, and alerting users when it’s back in stock.
Universal Cart Tools will be available to users in Search and the Gemini app this summer, followed by YouTube and Gmail. (AP)
(Except for the headline, this story has not been edited by federal officials and is auto-published from a syndicated feed.)
