LONDON, 19 May 2026 — Google has unveiled Gemini Spark, a 24/7 personal AI agent that marks a fundamental shift in how artificial intelligence integrates into daily life and professional workflows. Unlike conventional AI assistants that wait passively for user prompts, Spark operates autonomously in the background — executing multi-step tasks, learning individual working styles, and running scheduled automations even when a user's devices are switched off.

The launch positions Google at the vanguard of the emerging agentic AI race, where competition is no longer about which assistant can answer a question fastest, but which can reliably get things done without constant human supervision. Spark is currently rolling out to trusted testers and is available to Google AI Ultra subscribers over 18 in the United States, along with select business users, with broader availability expanding over coming weeks.

Key Takeaways

• Gemini Spark operates as a true autonomous AI agent, executing tasks in the background 24/7 — including when devices are powered off.
• It runs on Google DeepMind's Gemini 3.5 Flash and Antigravity models.
• Three core capabilities — Tasks, Skills, and Schedules — form the foundation of Spark's agentic architecture.
• Native integrations span Gmail, Google Calendar, Drive, Docs, Sheets, Slides, YouTube, and Google Maps, all disabled by default for user control.
• Spark is initially available to Google AI Ultra subscribers in the US, with business and international expansion planned for coming weeks.

What Spark Does: Tasks, Skills, and Schedules

Gemini Spark is structured around three interlocking capabilities that together enable a genuinely new class of delegated digital work. Understanding these three pillars is essential to appreciating why this launch is architecturally different from previous AI assistant updates — and why it represents a material step change in the consumer AI landscape.

Tasks

Tasks are the primary interface through which users delegate complex, multi-step work to Spark. Rather than asking a simple question, a user can issue a directive such as: "Help me find and track interior design internships in New Orleans for this summer." Spark then connects to the user's Google Workspace ecosystem — scanning Gmail, cross-referencing Google Calendar, and browsing the live web — executing the research autonomously from start to finish. The user can watch it work in real time or leave it running entirely in the background.

This crosses a meaningful technical threshold. Previous AI assistants, including earlier Gemini versions, were fundamentally reactive — they required a human to remain in the loop to advance each step. Spark breaks that dependency, more closely resembling a capable junior analyst than a search engine or chatbot.

Skills

Skills introduce a layer of personalisation unprecedented in consumer AI products. A user can instruct Spark: "Read through the last 50 emails I wrote and turn it into a style guide for how I write emails. Turn that into a skill called ghostwriter." From that point forward, every email drafted by Spark will match the user's authentic voice, tone, and formatting preferences — without repeated prompting.

This mechanism allows Spark to accumulate institutional knowledge about an individual, building a library of personal workflows that compound in value over time. Skills can be created for any repeated task: client onboarding, expense categorisation, competitive research, or content drafting. According to Google's AI blog, Skills are designed to eliminate repetitive prompting entirely, shifting the relationship between user and AI from transactional to collaborative.

Schedules

Schedules complete the agentic loop by enabling time-based and conditional triggers. A user can configure: "Every Monday at 9:00 AM, scan my inbox and review my emails from the past week. Give me a quick recap of the most important updates and provide a suggested, prioritised to-do list. Also schedule some calendar blocks for deep work." Spark executes this automatically — no manual initiation required.

Conditional triggers extend this further. For example: "When I receive an email enquiring about my photography services, automatically extract the client's name and requested date, log the lead in my Client Tracker Sheet, then create a new Google Drive folder named after the client." This combination of Tasks, Skills, and Schedules creates a system where a user's digital environment becomes increasingly automated and self-managing over time — without sacrificing user control or transparency.

---

Gemini Spark Core Capabilities

Feature	Description	Key Supported Apps	User Control
Tasks	Multi-step autonomous research and execution across apps and the open web	Gmail, Drive, Docs, Sheets, Maps, Web	Watch live or run silently in background
Skills	Personalised workflows trained on user behaviour and past data	Gmail, Docs, Sheets, Slides	User-defined, named and re-invocable at any time
Schedules	Time-based and conditional automated triggers requiring no manual initiation	Gmail, Calendar, Sheets, Google Keep	User-set triggers, paused or deleted at any time
Personal Intelligence	Cross-app context awareness enabling joined-up action across digital life	All connected Google apps	Off by default; opt-in per app in settings
Live Web Browsing	Real-time research across multiple sites, comparison, and booking assistance	Open web plus all Google services	Triggered by explicit user task
Background Operation	Autonomous server-side execution that continues when phone and laptop are off	Server-side; all connected services	On/off toggle per individual task

Personal Intelligence: Connecting the Digital Ecosystem

Central to Spark's differentiation is a capability Google calls Personal Intelligence. When enabled — and Google is emphatic that all connections are turned off by default and require explicit user activation — Spark can join dots across a user's entire digital footprint: emails, calendar events, documents, spreadsheets, location history via Google Maps, and watch history on YouTube.

The practical applications demonstrated by Google illustrate both the ambition and the commercial potential. Spark can monitor a user's subscription newsletters, synthesise themes across dozens of publications, and deliver a weekly intelligence briefing on the topics that matter most to that individual. It can turn a chaotic email chain about a group holiday into a structured master plan — logging receipts into a Google Sheets tracker and sending a co-ordinated group email on the user's behalf. For homeowners, it can analyse historical invoices to anticipate maintenance needs, set recurring Calendar reminders, and maintain a live supply list in Google Keep.

The browse-and-book capability is particularly significant for e-commerce and travel verticals. Spark can conduct live web research across multiple sites simultaneously, compare options, and in certain contexts complete bookings — a function that directly challenges purpose-built tools like Perplexity's shopping assistant and OpenAI's memory-enhanced agents.

As reported by TechCrunch, the agentic AI market is rapidly maturing, with enterprise and consumer use cases converging around the same core requirements: reliability, transparency, and deep app integration. Reuters has noted that Google's move follows a broader industry pattern, with Anthropic, Microsoft, and OpenAI all accelerating their own agent frameworks through 2026. According to Gartner's strategic technology trend research, agentic AI is among the highest-impact enterprise technology shifts through 2028, with adoption expected to scale sharply as trust and reliability benchmarks are established.

Our own coverage of the agentic sector has tracked this acceleration closely. The emergence of Spark is consistent with what we observed when NVIDIA's Hermes Agent attracted 140,000 GitHub stars in a matter of weeks — developer and enterprise appetite for capable autonomous agents is outpacing supply. Similarly, our analysis of Mistral's approach to 24/7 cloud-based agents identified a structural shift toward always-on AI infrastructure that Spark now accelerates at consumer scale.

---

Competitive Landscape: How Spark Compares

Product	Provider	Background Operation	Native App Integrations	Personalisation	Availability
Gemini Spark	Google	Yes — runs when devices off	Gmail, Calendar, Drive, Docs, Sheets, Maps, YouTube	User-defined named Skills	US, AI Ultra subscribers
Microsoft Copilot Agent	Microsoft	Partial — M365 background tasks	Outlook, Teams, Word, Excel, OneDrive	Limited personalisation	M365 Business subscribers
Claude Computer Use	Anthropic	No — session-based execution	Browser and desktop via API	Via system prompts	API access, developers
ChatGPT Tasks	OpenAI	Partial — scheduled reminders only	Limited via plugins	Memory features	Plus and Pro subscribers
Hermes Agent	NVIDIA	Yes — local model execution	Developer-configured	Fine-tuning required	Open source, developers

Why This Matters for Industry Stakeholders

For enterprise technology buyers, Gemini Spark raises immediate questions about workflow automation strategy. Organisations that have built tooling around Robotic Process Automation platforms will need to evaluate whether Spark's natural-language, agentic approach can deliver comparable reliability at significantly lower implementation cost and without specialised technical teams.

For the productivity software market, Spark signals a profound structural challenge to standalone tools. If a user's AI agent can organise files in Drive, synthesise research into Docs, and manage communications through Gmail autonomously, the marginal value of single-purpose productivity applications narrows considerably. As the Financial Times has noted, the platform battle in enterprise software is increasingly fought at the agent layer, not the application layer.

For investors, the launch reinforces Alphabet's position as a vertically integrated AI platform — controlling the model (Google DeepMind), the infrastructure (Google Cloud Vertex AI), the application suite (Workspace), and now the agent orchestration layer. This vertical integration is difficult for rivals to replicate and creates compounding competitive advantages in data access and personalisation depth that competitors operating at only one layer of the stack cannot match.

For consumers, the critical factor is trust. Spark is built with explicit guardrails: all app connections are opt-in and off by default, it checks with users before taking major actions, and it does not read emails indiscriminately — operating only within the specific scope a user has defined. As reported by Bloomberg, user trust in autonomous AI agents will be the defining adoption variable through 2026 and 2027. Related reading: our reporting on Netomi's $110M Series C and Anthropic's financial AI agents provides further context on how enterprise-grade agentic systems are being funded and deployed at scale in 2026. Figures independently verified via public financial disclosures and third-party market research.

Forward Outlook

Google has indicated that access to Spark will expand to more users and businesses over coming weeks. The international roll-out beyond the United States, and potential pricing changes as agentic capabilities move from the Ultra-tier premium to broader subscription tiers, will be critical signals to watch. The integration of Spark with Google's broader AI ecosystem and availability across Gemini's mobile and desktop clients is expected to expand the user base significantly once the trust framework is established at scale.

The deeper strategic question is whether Spark's 24/7 autonomous operation will prove reliable enough at enterprise scale to shift procurement decisions away from established automation platforms. Our head-to-head analysis of autonomous agent frameworks found that reliability under edge cases — not raw capability — remains the primary barrier to enterprise adoption. Spark will face exactly the same scrutiny as it scales beyond its initial trusted-tester cohort.

BUSINESS 2.0 has no commercial relationship with any companies mentioned in this article.

---

Bibliography

• Google — Gemini Spark: Your 24/7 Personal AI Agent (Official Product Page)
• Google — Gemini Overview
• Google AI — Official Site
• Google — The Keyword: AI Coverage
• Google DeepMind — Official Site
• Google — AI Ultra Subscription Details
• Google Workspace — Official Site
• Google Cloud — Vertex AI Platform
• Google — Gemini Help Centre
• Alphabet Inc. — Investor Relations
• Bloomberg — AI and Technology News
• Reuters — Technology Section
• Financial Times — Technology
• TechCrunch — Artificial Intelligence
• Gartner — Top 10 Strategic Technology Trends
• Anthropic — News and Research
• OpenAI — Memory and New Controls for ChatGPT
• Crunchbase — Google Company Profile
• YouTube (Google)
• Google Maps

Sources include company disclosures, regulatory filings, analyst reports, and industry briefings.

Related Coverage

• More Agentic AI Analysis
• Latest Technology News