Free Intensive Nov 21–22Free 2-day hands-on Vibecoding Intensive on Nov 21–22 at 16:00 MSK (Sat–Sun) · Live Cohort 3 starts Nov 23Join Free →Cohort 3 (Nov 23)
ReleaseInsight · On RadarThe Verge→ original

Google made its Spark AI agent free and renamed it Tasks in six countries

Google has renamed its Spark AI agent to Tasks and made it free. Previously it was available only to Google AI Pro and Ultra subscribers. The "Tasks" switch in Gemini is now available in six countries: the US, Canada, Japan, Korea, Australia and India. The agent builds meal plans and searches email for forgotten recurring payments. For product managers, this is a signal that agents are moving from paid tiers into mass-market products.

AI-processed from The Verge; edited by Hamidun News
Google made its Spark AI agent free and renamed it Tasks in six countries
Source: The Verge. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceThe Verge ↗
Editor-in-Chief
Key metric #1In the US, Tasks calls stores, compares prices and checks availability
Key metric #2Before a purchase made through Google Pay, the agent stops and waits for permission
Key metric #3Tasks run in a hardware-isolated virtual machine for each account

Google has renamed its AI agent Spark to Tasks and made it free. Spark is an assistant that does not just answer in chat but carries out routine work on its own: it plans meals, sorts through the calendar and searches email for forgotten subscriptions. Previously it was available only to people paying for Google AI Pro and Ultra plans. Now it has become a "Tasks" mode in the Gemini app and is open to all users in six countries: the US, Canada, Japan, Korea, Australia and India. This is one of the first cases where a full-fledged agent, rather than a regular chatbot, reaches a mass audience without a subscription.

What changed

Tasks is easy to find: on the Gemini home page, switch from "Chat" to "Tasks." The agent runs on a new model, Gemini 4 Hydrogen.

The basic set of tasks is fairly mundane: draft a weekly meal plan, tidy up the calendar, go through email and find recurring payments you may have forgotten. For example, a subscription to a service you haven't used for six months. You state the goal rather than dictating every step.

What is available only in the US

The most notable features are so far limited to the American market. The agent can call companies on the user's behalf to compare prices for a product or find out whether it is in stock. This removes the tedious chore of phoning five stores to track down one item.

The agent can also buy goods through Google Pay and sign into accounts on websites that support Google Password Manager. Tasks can be assigned without opening the app: you can write to [email protected] or send the agent a text message. In effect, this is an assistant you can message the way you would in a messenger app, not just a conversation partner in a chat window.

Security and control

The more the agent can do, the costlier its mistakes, and Google is betting on isolation and confirmations. According to the company, Gemini carries out tasks in a virtual machine with hardware isolation. It is dedicated to a specific account and is not shared with other users. This matters because the agent signs into accounts and works with email.

While a task is running, users can give the agent additional instructions, so there is no need to wait until it finishes to correct its course. Before sensitive actions, such as a purchase, the agent stops and asks for permission. This "the agent prepares, the human clicks confirm" pattern is becoming the standard for agents that handle money.

What this means for developers and product managers

First, the market for consumer agents is shifting from paid premium features to a free mass-market mode. If you are building your own agent for personal tasks, you will now compete not with a niche subscription but with a free option inside Gemini, tied to Google's email, calendar and payments.

Second, the set of channels is telling. Email and SMS as ways to assign tasks mean an agent's interface does not have to be a chat. For products with long-running tasks, this is a useful benchmark: the user gives an instruction, walks away, and gets the result later.

Third, step-by-step confirmation before payments is better built into the architecture from the start than added after the first user complaints.

There are also caveats. The material contains no data on how well the agent performs: how often it succeeds with purchases and calls, what usage limits Tasks has, or how it behaves in complex cases. There are also no timelines for when calls, payments and account sign-ins will appear outside the US. So for now this is more a bid for a new standard than a proven tool. But the move itself is clear: Google is turning the agent into an ordinary button in Gemini rather than a separate premium experiment.

Story

Key Questions Answered

What changed?

Tasks is easy to find: on the Gemini home page, switch from "Chat" to "Tasks." The agent runs on a new model, Gemini 4 Hydrogen. The basic set of tasks is fairly mundane: draft a weekly meal plan, tidy up the calendar, go through email and find recurring payments you may have forgotten. For example, a subscription to a service you haven't used for six months. You state the goal rather than dictating every step.

What is available only in the US?

The most notable features are so far limited to the American market. The agent can call companies on the user's behalf to compare prices for a product or find out whether it is in stock. This removes the tedious chore of phoning five stores to track down one item. The agent can also buy goods through Google Pay and sign into accounts on websites that support Google Password Manager. Tasks can be assigned without opening the app: you can write to [email protected] or send the agent a text message.

What are the key details on Security and control?

The more the agent can do, the costlier its mistakes, and Google is betting on isolation and confirmations. According to the company, Gemini carries out tasks in a virtual machine with hardware isolation. It is dedicated to a specific account and is not shared with other users. This matters because the agent signs into accounts and works with email. While a task is running, users can give the agent additional instructions, so there is no need to wait until it finishes to correct its course.

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Unified AI Workspace · Nomad

Test Frontier LLMs & Autonomous Agents in One Workspace

Run Claude Opus 5.5, GPT-6, Gemini 3.1, and DeepSeek side-by-side in Nomad AI (nomadnet.ai) — unified model routing, deep research, and multimodal workflows under one subscription.

Try Nomad AI →Launch Nomad Agent
What do you think?
Loading comments…