Subscribe
AI
2 min read

Google Turns Its Spark Agent Into Tasks and Makes It Free in Six Countries

Google on Thursday folded its Spark AI agent into the main Gemini app as a feature now called "Tasks," dropped the subscriber-only gate and widened availability to six countries. The move turns what was a premium capability, introduced back in May for Google AI Pro and Ultra subscribers, into the default agent layer for consumer Gemini. Tasks runs on Google's Gemini 4 Hydrogen model and handles the consumer-agent basics: building meal plans, managing calendars, and digging through email to surf

Olivetti brochure style cover: AGENTS GO FREE
Olivetti brochure style cover: AGENTS GO FREE

Google on Thursday folded its Spark AI agent into the main Gemini app as a feature now called "Tasks," dropped the subscriber-only gate and widened availability to six countries. The move turns what was a premium capability, introduced back in May for Google AI Pro and Ultra subscribers, into the default agent layer for consumer Gemini.

Tasks runs on Google's Gemini 4 Hydrogen model and handles the consumer-agent basics: building meal plans, managing calendars, and digging through email to surface recurring expenses worth cancelling. Google says the agent works inside a hardware-isolated virtual machine dedicated to a single account and never shared with other users, and that for sensitive steps, such as completing a purchase, it pauses and asks for permission first.

The American features are the aggressive ones

US users get capabilities that go further than scheduling:

  • Tasks can call businesses on the user's behalf to compare product pricing and check inventory.
  • It can purchase items using Google Pay.
  • It can sign into accounts that support Google Password Manager.
  • Users can assign tasks by emailing or texting gemini@google.com.

Availability outside the US covers Canada, Japan, Korea, Australia and India, where the feature set stops at the task-management layer. Inside the Gemini app, users switch the homepage toggle from Chat to Tasks.

Why it matters

Agentic assistants are crowded: Anthropic ships Claude Duet, OpenAI has Dots, and Meta is pushing Muse through WhatsApp. Google's differentiator was never the agent itself but the surface area it sits on, namely Gmail, Calendar, Android, and now a free tier that removes the subscription friction its rivals still charge for.

The call-businesses-and-buy-with-Google-Pay combination is the part to watch. An agent that phones a store to check stock and then completes the transaction over a consumer payment rail is the closest thing any major platform has shipped to a closed-loop consumer purchase, and Google Pay involvement means the flow now carries the trust, liability and fraud surface of a payment system. Regulators in the EU and US have spent 2026 examining exactly this class of agent-mediated transactions.

Making it free also solves a distribution problem every agent vendor has: autonomous agents that sit behind premium paywalls generate minimal usage data, and minimal usage data makes agent behavior hard to improve and hard to defend. Google is trading margin for telemetry and default status.

Gemini Tasks calling businesses and completing purchases with Google Pay
Each Gemini task runs in a hardware-isolated virtual machine

Sources

AIGoogleGemini