Vivold Consulting

Google's Gemini Spark brings always-on, cloud-run agentic help to consumers

Key Insights

Google unveiled Gemini Spark, a personal AI agent inside the Gemini app that takes action on a user's behalf and runs 24/7 on dedicated Google Cloud virtual machines, so it works in the background without a laptop open. Powered by Gemini 3.5 and Google's Antigravity harness, it handles long-horizon tasks and connects to third-party tools via MCP. It began rolling out to trusted testers, with a beta reaching Google AI Ultra subscribers in the US shortly after.

Stay Updated

Get the latest insights delivered to your inbox

Agentic AI aimed squarely at consumers

After bringing agents to developers and enterprises, Google used I/O to push the idea to everyday users with Gemini Spark, a personal AI agent in the Gemini app that navigates your digital life and takes action under your direction.

How it works

- It runs on dedicated virtual machines on Google Cloud and operates 24/7, so tasks continue in the background without keeping a device open.
- It's powered by Gemini 3.5 and Google's Antigravity harness, which lets it carry out long-horizon tasks.
- It integrates with tools - starting with Google's own and, in the following weeks, third-party tools via MCP (the Model Context Protocol).
- You can work with it in the Gemini app and, soon, through email and chat; on Android, a new UI space called Android Halo will show live agent progress, and later in the summer Spark will operate inside Chrome as an agentic browser.

The rollout

Google began rolling Spark out to trusted testers that week, with a beta coming to Google AI Ultra subscribers in the US the following week. It's part of a broader agentic push across Google's products - including a "Daily Brief" agent that synthesizes inbox, calendar, and tasks into a morning digest, and agentic capabilities arriving in Search. Spark is Google's clearest statement yet that it wants the agent, running safely in the cloud, to become a default way consumers get things done.

Related Articles

A US export order pulled Anthropic's top models offline worldwide, igniting an AI-sovereignty backlash

A US export-control directive forced Anthropic to abruptly disable Fable 5 and Mythos 5 for all foreign nationals on June 13, just four days after launch - briefly cutting off even its own overseas staff. Washington cited a jailbreak vulnerability; Anthropic disputed its severity but had to pull global access because it couldn't filter users by nationality in real time. Europe and Canada reacted with alarm, treating it as proof that frontier-AI access can be switched off by a single government overnight.

Huawei's agent-native HarmonyOS 7 moves into the China AI gap Apple can't fill

Four days after Apple confirmed Siri AI won't launch in China, Huawei unveiled HarmonyOS 7, restructuring the OS around an agent-native architecture it calls the beginning of the agent era. Its assistant Xiaoyi, rebuilt as a system-level agent, now drives 2,100+ system capabilities and coordinates 2,000+ third-party AI agents, atop the upgraded openPangu foundation model. With HarmonyOS already past iOS in China's smartphone share, independence forced by US sanctions has become a structural advantage in the one market Apple can't reach at the AI level.

US government orders Anthropic to pull its most powerful models, citing national security

The US government issued an export-control directive forcing Anthropic to immediately disable Fable 5 and Mythos 5 for all customers, citing national security and a reported jailbreak. Anthropic is complying but disputes the basis, arguing the cited technique surfaces only minor, already-known vulnerabilities that rival models can find without any bypass. Every other Claude model remains unaffected and available.