The most interesting thing about Google’s Tasks feature isn’t what it does. It’s who gets it. Agentic assistants have mostly been a premium upsell, and Google is now handing one to hundreds of millions of people at no cost.
Google announced the wider rollout of Tasks, a toggle inside Gemini that switches the app from Chat to a more active mode. Chat answers questions. Tasks does chores. It can call businesses and book appointments, complete purchases through Google Pay, and sign into websites securely when prompted. You can flip the toggle in the app, or text or call Gemini to give it instructions.
Josh Woodward, VP of Gemini, said in a blog post that he used it to phone local car dealerships and schedule a service appointment for his family’s minivan. He also had it analyze his electrical bill. Those are small, boring jobs, and that’s the point. Nobody wants an AI to run a company. They want it to sit on hold for them.
The history matters here. Tasks began in May as Gemini Spark, a 24/7 agentic helper offered to select testers, then widened to U.S. subscribers of the Google AI Ultra tier. It was later renamed Task Mode. Now it’s free for people using the new Gemini 4 Hydrogen model. That’s a fast slide from expensive experiment to default feature.
This puts pressure on the competition. OpenAI has pushed its own agent features in ChatGPT, and Anthropic has been building computer-use tools, but both have kept the most capable versions behind paid plans or developer access. Apple’s Siri still hasn’t delivered on the same promise. Google has something none of them can easily copy: Google Pay, a huge base of Android and Workspace users, and years of experience with phone-calling AI from Duplex.
Still, a few questions are open. Letting software spend money and log into accounts raises the cost of mistakes. A wrong answer in Chat is annoying. A wrong purchase or a botched booking costs real money. Google hasn’t said much in this announcement about spending limits, confirmation steps, or what happens when a business refuses to talk to an AI caller. Those details will decide whether people trust it.
My read: this isn’t noise. The feature set is modest, but distribution is everything in consumer AI, and free agents at this scale could reset what people expect from an assistant. Developers building narrow booking or errand apps should be nervous. Everyone else should wait to see how reliable it is before handing over a credit card.



