Gemini Spark Comes to macOS in Beta: Desktop File Automation, Remote Tasks, and New App Integrations
Gemini Spark is now in beta on macOS for US Google AI Ultra subscribers, bringing desktop file automation, remote task control, and new third-party integrations.
Gemini Spark launched at Google I/O in May as a cloud-based AI agent built to do real work on your behalf, not just answer questions. Since then it has lived primarily in the browser. As of June 30, 2026, that changes: Spark is now available in beta on the Gemini macOS app, and it brings a handful of updates that meaningfully expand what the agent can actually do.
The caveat up front: this is US-only, beta-only, and requires a Google AI Ultra subscription at $100 per month. If that is you, here is what you are getting.
Spark Can Now Work Directly with Files on Your Mac
The headline capability is desktop file automation. Spark can now reach into your local file system, with your permission, and do the kind of tedious file management that most people either put off indefinitely or spend a Saturday afternoon grinding through.
Practical examples from Google: ask Spark to sort all the PDFs in your Downloads folder into named subfolders, or pull the latest invoices saved locally and build a budget spreadsheet from them, complete with a recurring update schedule. That second one is worth pausing on, because it bridges your local files with Google Workspace. Spark does not just read your files; it connects what is on your Mac with Docs, Sheets, and the rest of the Workspace suite.
A new “Spark” tab appears in the sidebar of the macOS app (version 1.80.15.516 and above), mirroring the layout users already see on the web version.
Security matters here, and Google has been deliberate about it. Spark only accesses files you explicitly give it permission to use. Before taking anything high-stakes, like sending an email, making a calendar change, or spending money, it asks first. You are always in the loop on consequential actions.
Remote Task Assignment Is Coming Soon
One of the more interesting features is not live yet, but is worth knowing about. Google is preparing the ability to assign a task to Spark from your phone and have it execute on your Mac while you are away from your desk.
The practical scenario: you are out and remember you need to pull a specific revenue figure from a sales report on your Mac and email it to someone. You assign that to Spark from your phone, and it handles the whole chain on your computer while you carry on with your day.
Because Spark is cloud-based, it keeps running even when you close your laptop or lock your phone. The remote task feature has been spotted in Android app code under the internal codename “Robin,” with interface strings like “Select a computer” and “Make sure your Mac is online.” It uses an isolated thread system designed to keep data from that task separate from your other conversations with Gemini.
New App Integrations via MCP
Alongside the macOS launch, Google is rolling out new third-party integrations built on Model Context Protocol, the open standard originally created by Anthropic and now widely adopted across the AI tooling space.
The new connections include:
- Canva for designing flyers and graphics
- Dropbox for accessing and sharing files
- Google Tasks and Google Keep for personal organisation (a useful pairing: Spark can scan your scattered Keep notes and convert them into structured Tasks)
- OpenTable for restaurant reservations
- Instacart for grocery ordering
- Zillow Rentals for scheduling apartment tours
These integrations are rolling out over the next week on Gemini Spark on web and mobile. The macOS app will get them in the coming weeks.
If none of the named partners cover your needs, Google has also opened up custom MCP support. At the bottom of gemini.google.com/apps, you can enter a custom app link to connect additional services. Each connected app is exposed as an MCP server, and Spark calls it through a sandboxed runtime. Your credentials are never passed to the language model itself.
Every app connection has to be explicitly enabled in settings. Spark does not touch anything you have not authorised.
Real-Time Topic Tracking
One more addition rolling out alongside the macOS launch: Spark can now monitor topics and react to events in real time. You can point it at blogs, news sites, social media, finance feeds, sports results, or weather, in addition to the Gmail monitoring it already supported.
Some examples of how this plays out: ask Spark to send you a detailed financial summary when a specific stock hits a target price, or deliver match highlights the moment a game ends. These are scheduled, event-driven tasks, which fits the way Spark is structured: every job has a task (what to do), a schedule (when to do it), and optionally a skill (reusable instructions for how to do it). You can have up to 15 tasks running concurrently.
What to Expect Next
Google has been clear that this macOS release is an early step. Voice interaction for the Mac app was announced at I/O but is not included here. Chrome integration, which would let Spark take actions directly in the browser, is planned for later this year. Android users will also eventually get a new UI layer called Android Halo to monitor live task progress.
The big picture: Spark is still very much a work in progress being rolled out in stages. If you are a Google AI Ultra subscriber in the US with a Mac, it is worth trying. The file automation alone could save a meaningful amount of time if your Downloads folder looks anything like most people’s. For everyone else, the integrations coming to web and mobile are the more immediately accessible piece of this update, with no desktop required.