Google I/O highlights agentic AI

Google used its May 19 I/O conference to make “agentic” AI the central theme of its product push, tying new models, consumer features and developer tools to software that can carry out multi-step tasks. CEO Sundar Pichai titled his keynote “Welcome to the agentic Gemini era,” while Google DeepMind and product executives described Gemini 3.5 Flash, Gemini Spark and Antigravity as the core pieces of that strategy. (blog.google 1) (blog.google 2) Google’s announcements were spread across Search, the Gemini app, Google AI Studio and enterprise products, rather than presented as a single standalone agent product. (blog.google) That framing matched Google’s own description of the event as a move “from prompts to action,” and it was echoed in follow-up coverage from PPC Land on May 23. ### What did Google mean by “agentic” at I/O? Sundar Pichai said on May 19 that Google is now focused on showing “the value in the products” people use every day, after what he called a year of rapid shipping and AI advances. (blog.google) In Google’s I/O materials, “agentic” referred to systems that do more than answer questions — they plan, execute and manage tasks across steps and surfaces. Koray Kavukcuoglu, Jeff Dean, Oriol Vinyals and Noam Shazeer wrote that Gemini 3.5 was built to “execute complex, agentic workflows,” with the company saying the model is aimed at long-horizon tasks in coding and other work. (blog.google) Google’s developer post used similar language, saying I/O was accelerating a shift “from prompts to action.” ### Why was Gemini 3.5 Flash the centerpiece? Google said Gemini 3.5 Flash is the first release in the new 3.5 family and is now the default model in the Gemini app and AI Mode in Search globally. The company described it as its strongest agentic and coding model so far and said it outperformed Gemini 3.1 Pro on several coding and agentic benchmarks. (blog.google) The company also said 3.5 Flash runs at four times the speed of other frontier models when measured by output tokens per second. Google said that balance of speed and performance makes the model suitable for long-horizon tasks, including application development, code maintenance and financial document preparation. (blog.google) ### What is Gemini Spark supposed to do? Josh Woodward, Google’s vice president for Google Labs, the Gemini app and AI Studio, said Gemini Spark is a “24/7 personal AI agent” designed to proactively manage tasks under the user’s direction. Google paired that with a new “Daily Brief” agent that gives users a personalized morning update. (blog.google) Google also said the Gemini app now reaches more than 900 million people each month across 230 countries and more than 70 languages, up from 400 million at the time of last year’s I/O. The company said Spark will also be integrated into its macOS app so it can operate on a local machine and add new voice features. (blog.google) ### Where does Antigravity fit in? Varun Mohan and Logan Kilpatrick of Google DeepMind said Antigravity is Google’s “agent-first development platform” for turning an idea into a production-ready application. At I/O, Google introduced Antigravity 2.0 as a standalone desktop app, alongside an Antigravity CLI and SDK. (blog.google) Google said the platform is built to orchestrate multiple agents in parallel, support scheduled background tasks and connect with AI Studio, Android, Firebase and Google Cloud. In the Gemini 3.5 post, the company said an updated Antigravity harness lets 3.5 Flash deploy collaborative subagents for multi-step workflows under supervision. (blog.google) ### What comes next after I/O? Google said Gemini 3.5 Pro is already being used internally and is planned for release next month. The company also said Gemini Spark integration for the macOS app is coming, while Antigravity 2.0 and the new AI Ultra plan at $100 a month were introduced as part of the post-I/O rollout. (blog.google) (blog.google)