Google's auto browse, the Chrome agent that carries out tasks across the open web, starts reaching Android phones at the end of June. Built on Gemini 3.1, it can do things like reserve parking from a ticket confirmation through SpotHero or update a recurring Chewy order on request.
The fine print is where the ambition meets reality. Auto browse on Android is limited to AI Pro and Ultra subscribers in the US, on devices running Android 12 or higher with at least 4GB of RAM and the language set to English-US. Google also builds in a brake: the agent is designed to ask for confirmation before sensitive actions like making a purchase or posting to social media, and it carries the same prompt-injection defences as the desktop version.
Putting an agent inside the default mobile browser is a different distribution game from a standalone app. The subscription gate and the confirmation step show Google is still feeling out how much autonomy users will tolerate on their own devices.