Google centers Gemini 3.5 Flash in new enterprise agentic‑workflow push at I/O

Google used its May 19 I/O conference to make a narrower, more operational case for “agentic AI” than the chatbot framing that has dominated the past two years. The company centered Gemini 3.5 Flash as the model for long-running, multi-step work across Search, developer tools and enterprise software, and paired it with new orchestration products designed to keep tasks running in the background. Google also introduced Gemini Spark, which it described as a 24/7 agent that can proactively manage tasks under a user’s direction. ### Why did Google put Gemini 3.5 Flash at the center of the pitch? Google said on May 19 that Gemini 3.5 Flash is the first model in a new family that combines “frontier intelligence with action,” and it described the model as built for complex agentic workflows rather than one-shot answers. In Google’s product rollout, 3.5 Flash is available in the Gemini app, AI Mode in Search, Google Antigravity, the Gemini API, Android Studio, Gemini Enterprise Agent Platform and Gemini Enterprise. (blog.google) Google executives said speed was part of the argument. In its model announcement, the company said 3.5 Flash is “4 times faster than other frontier models” on output tokens per second and outperforms Gemini 3.1 Pro on several coding and agentic benchmarks, including Terminal-Bench 2.1, GDPval-AA and MCP Atlas. ### What does Google mean by “agentic workflows” here? (blog.google) Google’s own language at I/O moved from assistance to execution. The company said 3.5 Flash is suited to “long-horizon agentic tasks,” and said that, when paired with the updated Antigravity harness, it can deploy collaborative subagents, execute multi-step workflows and sustain performance under supervision. (blog.google) Google’s developer briefing made that shift more explicit. Antigravity 2.0, the company said, is designed as a central home for agent interaction, where multiple agents can be orchestrated in parallel, with dynamic subagents for parallelized workflows and scheduled tasks for background automation. ### Where does Spark fit in? (blog.google) Google said Gemini Spark is a “24/7 personal AI agent” that is meant to manage tasks continuously rather than respond only when prompted. In the Gemini app update, Josh Woodward, vice president of Google Labs, said Spark is designed to help users “get things done around the clock,” and Google said the MacOS app will integrate Spark so it can operate on a local machine as well. (blog.google) Outside coverage of the launch described Spark in similar terms. The Next Web reported that Google presented Spark as an always-on agentic assistant that can be emailed like a colleague and run continuously on cloud virtual machines for AI Ultra subscribers. ### How much of this was aimed at enterprises, not consumers? Sundar Pichai used the keynote to tie Google’s product announcements to enterprise demand. (blog.google) He said Google is now processing more than 3.2 quadrillion tokens per month across its surfaces, that more than 8.5 million developers are building with its models monthly, and that more than 375 Google Cloud customers each processed more than one trillion tokens over the past 12 months. (thenextweb.com) Google also placed enterprise distribution around the model from the start. The 3.5 Flash launch page said the model was available for enterprises in Gemini Enterprise Agent Platform and Gemini Enterprise, while the developer update said Antigravity can connect directly to Google Cloud projects to simplify enterprise workloads. ### What changes if these tools work as advertised? (blog.google) Google’s examples focused on process work rather than conversation. The company said 3.5 Flash can help complete tasks that “used to take a developer days or an auditor weeks,” including developing applications, maintaining codebases and preparing financial documents, and said the model can plan, build and iterate across real-world problems. (blog.google) That framing points toward broader automation of data gathering, coding, testing and production handoffs, with humans supervising the workflow rather than driving each step manually. CIO, in its coverage of I/O, said Google repeatedly emphasized agents that can reason, act and orchestrate workflows with minimal human input for enterprise use cases. (blog.google) ### What comes next from Google’s rollout? Google said Gemini 3.5 Pro is already being used internally and is scheduled to roll out next month. On May 19, the company also launched Antigravity 2.0, Managed Agents in the Gemini API and a new $100-per-month AI Ultra plan with higher Antigravity usage limits and bonus credits for subscribers who hit quota caps. (blog.google) (cio.com)