Google is set to introduce various new AI-powered tools and systems. Among these is an AI assistant designed to perform tasks for users proactively. This innovation, referred to as “agentic” AI, was highlighted at Google’s annual developers conference, Google I/O. One of the key announcements was the AI agent named Gemini Spark. “We are firmly in our agentic Gemini era,” stated Google CEO Sundar Pichai during the event at the company’s Mountain View, California headquarters. He added that while there has been significant progress, making these agents user-friendly and secure remains a work in progress.
Google and its parent company, Alphabet Inc., have invested billions into AI development. In a recent call with investors, Google’s top finance executive noted that capital expenditures for this year could reach up to $190 billion. According to their latest earnings report, these investments are yielding positive returns with notable growth, and Google’s stock has risen by 11% since the report.
Gemini Model Advancements
Last year, the Gemini app had 400 million monthly active users; current usership has more than doubled, now exceeding 900 million. The latest Gemini version, Gemini 3.5, was introduced on Tuesday. This model, including Gemini 3.5 Flash, emphasizes speed and offers enhanced coding capabilities. Google claims it is four times faster than some rivals. This model is set as the default for the Gemini app and “AI mode” in Google search. The company also mentioned that Gemini Pro’s 3.5 version is in the works and anticipated for release next month.
Gemini 3.5 was developed with improved safety training to minimize the creation of harmful content or erroneous refusal to answer safe queries, as per Google. Furthermore, Google announced a new offering, Gemini Omni, allowing users to create high-quality videos using any input—text, images, videos, and audio. Although users will eventually be able to create images and audio, Google did not specify a rollout date for these features. Omni’s videos will have a realistic appearance, benefiting from an understanding of elements such as gravity and kinetic energy.
The initial model, Gemini Omni Flash, launches Tuesday for Google AI Plus, Pro, and Ultra subscribers. Starting this week, it will be accessible at no charge via YouTube Shorts and YouTube Create App. An imperceptible digital watermark, SynthID, will accompany all videos made with Omni. Additionally, content credentials verification in the Gemini app will help determine the creation source of content such as photos or videos.
24/7 AI Agent and Wearable Technology
Gemini Spark, powered by Gemini 3.5, will handle routine tasks like managing meeting notes, emails, and chats, as well as compiling takeaways into a document. Unlike other agents, Spark operates in the cloud, allowing uninterrupted background work even when devices are off. Though proactive in nature, Spark requires user permission for “high-stakes” activities like sending emails or making purchases. Select testers will have access starting Tuesday, with a broader rollout planned for Google AI Ultra tier subscribers in the U.S. This summer, Spark will integrate into Chrome.
The conference also updated attendees on the eagerly awaited Google smart glasses. Two versions are in development: audio glasses offering spoken assistance and display glasses providing visual information. The audio glasses are expected later this fall. Users can activate Gemini by saying “Hey Google” or tapping the frame, with functionalities like navigation support and real-time translations.
Google partnered with Samsung and eyewear brands Gentle Monster and Warby Parker for these glasses, showcasing designs at the conference. These are set to launch alongside the full eyewear collections later this year.
AI Enhancements in Search and Shopping
The previous year’s conference spotlighted “AI mode” in Google’s search engine, which presents a more conversational response before displaying relevant links. AI mode queries have more than doubled each quarter since its introduction, recently reaching over 1 billion monthly users, according to Liz Reid, head of Google search. The default search model is now Gemini 3.5 Flash. A major upgrade, dubbed the intelligent search box, accommodates longer queries and aids users with AI-driven suggestions rather than traditional autocomplete methods. The search options now include text, images, videos, files, and even Chrome tabs.
A new feature, the Universal Cart, is described as an “intelligent shopping cart,” operational across different merchants and services. Users can add items while exploring Google search, conversing with Gemini, watching YouTube, or reading Gmail. This cart leverages Gemini models to seek deals and notify users of price changes or restocks. Available this summer, it will initially roll out on search and the Gemini app, with YouTube and Gmail following suit.
This article was contributed to by Associated Press Writer Barbara Ortutay in Oakland, California.
Leave a Reply