Gemini Spark Is Here: Google's 24/7 AI Agent — What It Can Do, How to Access It, and Why It Matters

Google made clear at I/O 2026 in May that it wants Gemini to become an “operating layer” for how people work — and the centerpiece of that vision is Gemini Spark.

Billed as a “24/7 personal AI agent,” Spark runs on dedicated Google Cloud VMs, stays active even when your phone is locked or your laptop is closed, and can execute multi-step tasks autonomously. It’s not a chatbot that waits for your question. It’s a persistent agent you direct once and let run.

The rollout timeline has moved fast from announcement to reality. Here’s the current state as of late July 2026.

What Is Gemini Spark?#

Gemini Spark is an always-on AI agent that lives in Google’s cloud infrastructure — not on your device. It’s powered by Gemini 3.5 Flash (Google’s latest frontier model that benchmarks ahead of GPT-5.5 and Claude Opus 4.7 on many agentic coding tasks) and built using the Antigravity agent harness.

Key architectural difference from traditional assistants: Spark operates asynchronously. You give it a task — “every weekday morning, email me a summary of urgent overnight messages” — and it executes on that schedule without you needing to be present, prompt it again, or keep the app open.

Source: Google Blog — The Gemini app becomes more agentic (May 19, 2026)

What Can It Actually Do?#

Spark’s capabilities have expanded significantly since its May announcement.

Core capabilities (available at launch)#

Gmail triage & drafting: Read, summarize, and draft emails. Set recurring schedules for inbox scans.
Google Calendar integration: Manage events, check availability, schedule meetings.
Google Docs, Sheets, Slides: Create documents from prompts, populate spreadsheets, build presentations.
Google Tasks & Keep: Turn scattered notes in Keep into structured action items in Tasks.
Proactive scheduled tasks: Set recurring agent tasks (e.g., “check for vendor invoices every Monday and put them in a spreadsheet”).

Source: Google Cloud Blog — Innovations from Google I/O 26 (May 19, 2026)

June 2026 updates#

macOS desktop app (Beta): Spark can now access local files on your Mac — sort PDFs, pull data from local invoices into Google Sheets, manage files. Remote execution coming soon: assign a task from your phone, and Spark runs it on your Mac while you’re away.
Third-party integrations: Canva (design creation), Instacart (grocery ordering), OpenTable (restaurant reservations), Dropbox (file access), Zillow Rentals (apartment tours), Google Tasks and Keep.
Custom MCP (Model Context Protocol): Connect your own apps into Spark through MCP. This is significant: it means Spark is extensible beyond Google’s ecosystem.
Real-time topic monitoring: Spark can track news, sports, finance, weather, social media, and email — and notify you when conditions are met (e.g., “tell me when this stock hits $X”).

Source: Google Blog — Gemini Spark updates: macOS launch, connected apps and more (June 30, 2026)

What’s coming next#

Google has confirmed future features including texting and emailing Spark directly (it gets its own contact), creating custom sub-agents, and operating your local browser. No firm dates yet.

Source: Google Blog — The Gemini app becomes more agentic

Pricing and Availability#

This is where the picture gets more nuanced.

Plan requirements#

Plan	Spark Access	Price
Free	❌ Not included	$0
Google AI Plus	❌ Not included	$7.99/mo
Google AI Pro	✅ Rolling out (US & India, English)	$19.99/mo
Google AI Ultra	✅ Available in supported countries	$99.99/mo (was$ 249.99)

Availability timeline#

May 19, 2026: Announced at I/O. Trusted testers get access.
May 25, 2026: Beta launches for US Google AI Ultra subscribers.
June 17, 2026: Expanded to all countries where Gemini Apps are supported (with some exclusions).
June 30, 2026: macOS desktop app beta launches.
July 14, 2026: Added Australia, Canada, Hong Kong, India, Japan, South Korea for Ultra users.
July 16, 2026: First major expansion — Gemini Spark begins rolling out to Google AI Pro subscribers in the US (English only). This dropped the entry price from $99.99 to$ 19.99/mo.
July 29, 2026: Expansion to Google AI Pro subscribers in India announced, rolling out over the next few weeks.

Currently excluded regions#

European Economic Area (EEA), Nigeria, Switzerland, United Kingdom. No launch dates announced.

Source: Google Support — What’s new for Gemini Spark | AI Agents Library — Availability tracker (July 29 update)

How It Differs from OpenAI’s ChatGPT or Claude#

Dimension	Gemini Spark	ChatGPT / Claude
Persistence	24/7 cloud-based, runs without device	Session-based, requires user prompt
Scheduled tasks	Native recurring schedules	No native scheduling
Proactive alerts	Yes — topic monitoring, threshold triggers	No — passive response only
Workspace integration	Deep Google Workspace (Gmail, Docs, Sheets, Calendar)	Via API/plugins, less native
File system access	macOS desktop app (local files)	Limited
MCP	Yes — custom MCP support	Limited/partner MCP
Price	$19.99–$ 99.99/mo	$20–$ 200/mo

The biggest differentiator is the proactive, scheduled execution model. Spark doesn’t need you to show up and ask. You define a recurring workflow once, and it runs independently. This is closer to how a human assistant operates than how a chatbot operates.

What This Means for Developers#

MCP integration is real#

Google committed to custom MCP support in Spark, which means third-party tool builders can extend Spark’s capabilities the same way they extend other MCP-compatible agents. This aligns Spark with the broader MCP ecosystem we’ve been writing about extensively on this blog.

Managed Agents API#

At I/O, Google also announced the Managed Agents API — a service that lets developers spin up custom agents with a single API call. Each agent gets its own ephemeral Linux sandbox with skills, MCP servers, and server-side tools. Powered by Antigravity and Gemini 3.5 Flash.

Source: Virtualization Review — Google I/O ‘26 Fills Out Enterprise Agent Stack (May 19, 2026)

ADK 2.0#

The Agent Development Kit 2.0 (open source, available at adk.dev) now supports Python, Node.js, Go, and Java. It’s a code-first framework for building multi-agent systems, with native deployment to Google Cloud.

The Bottom Line#

Gemini Spark marks a genuine architectural shift in consumer AI: from session-based chatbot to persistent, cloud-resident agent.

The rollout is happening fast — from 0 to 1+ million eligible users in two months. The Pro rollout (July 16) was a key milestone, dropping the barrier from $100 to$ 20/month in the US and India. If Google continues expanding, Spark could reach a significant portion of Gemini’s 900+ million monthly active users by the end of 2026.

For developers and AI practitioners, the intersection of Spark + Managed Agents + ADK 2.0 + MCP support creates a compelling platform story. Google is not just launching a product — it’s establishing a multi-tier agent ecosystem with low-code paths (Agent Studio), managed runtime (Managed Agents API), pro-code frameworks (ADK), and a consumer agent (Spark) all running on shared infrastructure.

The question that remains open: will users trust a 24/7 agent with access to their email, calendar, files, and third-party services? That’s the adoption bottleneck — not the technology.