Google has made its Spark AI agent available to free Gemini users and renamed the feature Tasks, opening access to an automated assistant that can take on jobs such as shopping, appointment booking and bill analysis. The change gives people a way to try an AI agent without subscribing to a paid plan, but access is limited, and the feature is not available in every country.

Tasks can be selected from the standard Gemini Chat setting. The rebrand from Spark appears to have rolled out gradually in mid-October, and Google has not indicated that the change represents a substantial shift in the agent’s capabilities. It is an agentic version of Gemini, distinct from the company’s existing Google Tasks to-do-list product, which lets users organize items such as tasks drawn from email.

Free access, with boundaries

Google’s offer is not unlimited use. The source report describes access as capped at five hours at a time, although the precise allowance available to free users is not clearly specified. The distinction matters because agentic systems may need to work through several steps, rather than simply answer a single question, and users may need to manage access when they want the agent to handle more involved requests.

There is some indication of how the cap may feel in practice, though it comes from a paid-plan test rather than a free account. During a trial of Spark included with Google AI Pro, ZDNET’s tester used 20% of the five-hour allowance while carrying out four substantial tests. That experience suggests some users may find the available time sufficient for occasional experimentation, but it does not establish how quickly free users will reach their limit or how the allowance is applied across different tasks.

The feature is available in the United States, Canada, South Korea, Japan, Australia and India. In the US, users can also initiate tasks by emailing or texting Gemini, and the agent can call businesses on their behalf. Those calling capabilities are unavailable in Indiana, Louisiana, Minnesota, Montana and Nebraska, according to the report.

Part of a crowded agent market

The free tier gives Tasks a different price point from several competing personal agents. OpenAI’s Dots agents are included with ChatGPT Pro, which costs $100 per month, while Anthropic’s Claude Duet is offered with paid plans starting at $20 a month. Meta has also positioned its Muse agent as free to use up to 100 million tokens per week, a limit Mark Zuckerberg cited at launch in September. However, Meta’s help information is less explicit about the allowance, making direct comparisons difficult.

Price is only one part of the comparison. ZDNET’s AI tester-in-residence had reservations about Spark after testing it, while early evaluations of Dots and Duet have also been mixed, as described in the source report. The available information does not establish that Tasks is more capable than those alternatives. Its immediate distinction is that users in supported markets can try it without paying for an AI subscription, subject to limits.

For businesses and individual users, that creates a relatively low-cost way to assess whether an agent can handle routine, multi-step work. Tasks is described as running on Google’s Gemini 4 Hydrogen model and can be asked to help with activities including grocery shopping, booking appointments and reviewing a heating bill. These examples point to a broad personal-assistance role, but they also involve varying levels of judgment, access to personal information and, in some cases, spending decisions. Google says users can set spending caps for orders.

Convenience brings privacy questions

Because an agent may act proactively, connect to third-party services or review files, granting it access raises privacy and security considerations beyond those of a conventional chat. Google says users can change task settings, disconnect Google and third-party apps, or delete chat history. The company has also introduced a Gemini Memory summary that lets users review and edit personal information the chatbot retains. Those controls offer ways to manage data and connections, but users still need to consider what information and permissions a task requires.

Google’s move follows the retirement of Gems, its customizable Gemini assistants, and the conversion of existing Gems into Skills. The Tasks landing page also emphasizes personal finance, presenting the agent as a tool for finding potential savings, such as identifying recurring expenses or tracking desired deals. For now, the free offer lowers the barrier to trying an agent, not the need to weigh its limits, reliability and access to personal data. How useful it becomes will depend on what users can accomplish within those limits and whether the convenience justifies the permissions involved.