Google is set to introduce a range of new artificial intelligence tools, including an AI assistant designed to proactively assist users. This concept, often termed as “agentic” AI, was highlighted during Google’s annual developers conference, Google I/O. Among the announcements was a new AI agent called Gemini Spark.

Gemini Spark: The Future of AI Assistants

Google CEO Sundar Pichai addressed attendees, emphasizing the transformative potential of AI agents, while noting the challenges of making them user-friendly and secure. The company’s significant investment in AI is evident, with plans to potentially spend up to $190 billion this year. This investment aligns with Google’s growth, as showcased in its recent financial results. The Gemini app reflects this success, with active users growing from 400 million to over 900 million in the past year.

Gemini 3.5: Speed and Safety

The newly introduced Gemini 3.5 model, debuting globally, offers improved speed and capabilities. The Flash model within Gemini 3.5 focuses on efficiency and is now integrated with Google Search’s AI mode. An upcoming 3.5 Pro version is also in development. Enhanced safety and mitigation features ensure Gemini 3.5 generates less harmful content and handles queries better.

Google also revealed Gemini Omni, a tool allowing users to create videos from diverse inputs like text and audio. These videos can then be edited interactively. While Omni’s capabilities will expand to include image and audio creation, no specific release dates were given. Omni’s understanding of physical forces ensures more realistic video output than other models.

Videos crafted with Omni will include Google’s SynthID digital watermark, with additional content verification features forthcoming in Chrome’s search functionality. Google collaborated with companies like OpenAI and Eleven Labs to expand SynthID use.

24/7 Cloud-Based AI: Gemini Spark

Gemini Spark, powered by Gemini 3.5, automates routine tasks such as sorting emails and compiling documents. Its cloud-based nature allows it to function continuously, even when devices are off. Unlike traditional chatbots, Spark requests permission for significant actions like purchasing or emailing.

Testers can access Spark, with beta testing for U.S.-based Google AI Ultra subscribers following. By summer, Gemini Spark will integrate with Chrome.

Innovative Wearable AI: Smart Glasses

Google announced new smart glasses at the conference. Audio glasses provide assistance via audio cues, while display glasses offer visual information. Users interact with the glasses using voice commands or gestures.

These glasses, developed with Samsung, Gentle Monster, and Warby Parker, will launch later this year as part of full eyewear collections.

Enhanced Search and Shopping Experience

Last year’s introduction of “AI mode” in Google Search allowed for more conversational user queries. This feature has rapidly grown, boasting over 1 billion monthly users. With the launch of Gemini 3.5 Flash, Google is enhancing the search box with capabilities that accommodate longer queries and suggest AI-driven question formats.

Google also introduced the Universal Cart, an innovative shopping tool. It allows users to integrate items from various platforms into one cart, seamlessly tracking deals and restocks. This functionality will debut in search and the Gemini app this summer, followed by YouTube and Gmail.

Associated Press Writer Barbara Ortutay contributed to this story from Oakland, California.