Google just handed everyone the same AI assistant that was reserved for paying customers, and it can spend your money.
The Summary
- Google made Gemini Tasks free in six countries, dropping the paywall on agent features that previously required a subscription
- Tasks can now make phone calls, log into websites, and purchase items on your behalf, crossing into actual transaction territory
- New U.S. features include deeper Google Workspace integration, letting the agent act across your existing tools and data
- The shift signals Google racing OpenAI and Anthropic to get personal AI agents into mass adoption first
The Signal
Google removed the subscription barrier to Gemini Tasks, making agent capabilities that could previously cost you a monthly fee available for free across six countries. This is not about being generous. This is about getting millions of people comfortable with AI agents doing real tasks before competitors can.
The feature set matters more than the pricing. Tasks can coordinate phone calls, handle website logins, and execute purchases. That last one is the threshold crossing. An AI that can read your email is a productivity tool. An AI that can spend your money is an economic agent.
"An AI that can read your email is a productivity tool. An AI that can spend your money is an economic agent."
The U.S. rollout adds Google Workspace integration. This means the agent can touch your calendar, your docs, your drive, your corporate email. It is not operating in a sandbox anymore. It is operating in your actual workflow, with access to the context that makes it useful: who you work with, what projects you are juggling, which meetings you are avoiding.
The timing tracks with the broader agent race. OpenAI demoed Operator. Anthropic shipped computer use in Claude. Now Google gives away what it was charging for a month ago. The pattern is clear:
- Phase 1: Charge early adopters premium prices for beta agent features
- Phase 2: Drop prices to zero to flood the market with users before competitors do
- Phase 3: Monetize the platform, the integrations, the enterprise tier, or the data
We are in phase 2. The land grab is on.
The Implication
If you have been waiting to test AI agents because of cost, the barrier just dropped. Start with low-stakes tasks: scheduling, research, data entry. See where it breaks. See where it surprises you. The companies that figure out agent workflows now will have six months on the ones who wait for the platforms to mature.
Watch how Google monetizes this. Free agent access does not stay free unless the business model is somewhere else. That somewhere is likely enterprise Workspace licenses, advertising tied to agent actions, or transaction fees when your agent books that flight. The agent is free. The economic layer it plugs into is not.