Google launched Gemini Spark, its 24/7 agentic AI assistant, on macOS on 1 July 2026 in a beta release added to the existing Gemini desktop app [1][2]. The release brings real-time topic tracking, connections to more apps such as Google Tasks and Google Keep, and custom Model Context Protocol support [1].

Model Context Protocol (MCP) is an open standard that lets third-party apps plug directly into an AI assistant, so users build an assistant tailored to the tools they use every day [1].

Why the Mac release matters

The macOS launch lets Gemini Spark compete with desktop AI agents like Claude Desktop, Microsoft Copilot, and OpenClaw [1]. Spark can now work with files on the computer, and later handle remote tasks [1].

Google says users will soon be able to assign multi-step tasks to Spark from their phones, such as calling the desktop agent to pull information from a file on their Mac [1]. Today Spark can sort and organize files, or use files on the computer as the source for a new Google Workspace doc or spreadsheet, for example turning invoices into a budgeting worksheet [1].

App integrations and real tasks

Spark now integrates with Google Tasks and Keep, a gap TechCrunch flagged when Spark launched in May [1][3]. Spark also connects to third-party apps including Canva, Dropbox, Instacart, OpenTable, and Zillow Rentals [1].

Those integrations let Spark perform tasks like reserving tables, ordering weekly groceries, designing flyers, or booking apartment tours [1].

Real-time tracking

Gemini Spark can now track topics and react to events in real time, improving performance for tasks like tracking sports scores, stock movements, or breaking news [1]. Spark can also watch other areas like social media, blogs, online shopping, and weather [1].

Availability

Gemini Spark for macOS in beta is currently available only to Google AI Ultra subscribers in the United States [1].

What it means for agent builders in Indonesia

For AI agent builders in Indonesia, three signals are worth noting. Custom MCP support opens a path to plug your own apps into a desktop assistant without building proprietary integrations [1]. The cross-device multi-step task pattern, a phone calling a desktop agent, points to agent architectures that span surfaces [1]. Tight desktop agent competition among Google, Anthropic, Microsoft, and OpenClaw is pushing interoperability standards like MCP to mature fast [1].

Frequently asked questions

What is Gemini Spark?

Google's 24/7 agentic AI assistant that can run multi-step tasks, work with files on the computer, and connect to apps like Google Tasks, Keep, Canva, and Dropbox. The macOS version launched 1 July 2026 [1][2].

What is custom MCP support in Gemini Spark?

Model Context Protocol is an open standard that lets third-party apps plug directly into Spark, so users build an assistant tailored to their own tools [1].

Where is Gemini Spark for Mac available?

The macOS beta is currently available only to Google AI Ultra subscribers in the United States as of 1 July 2026 [1].

What tasks can Gemini Spark run?

Sort and organize files, use files as the source for a Google Workspace doc or spreadsheet, reserve tables, order groceries, design flyers, book apartment tours, and track sports scores, stocks, and news in real time [1].

Sources

  • 1. TechCrunch, Sarah Perez, Gemini Spark, Google's agentic assistant, is now available on Mac. https://techcrunch.com/2026/07/01/gemini-spark-googles-agentic-assistant-is-now-available-on-mac/
  • 2. Google blog, Gemini Spark updates. https://blog.google/innovation-and-ai/products/gemini-app/gemini-spark-updates-june-2026/
  • 3. TechCrunch, I put Google's 24/7 AI assistant Gemini Spark to work. https://techcrunch.com/2026/05/30/i-put-googles-24-7-ai-assistant-gemini-spark-to-work-and-its-actually-pretty-useful/